Merge remote-tracking branch 'lucko/master' into forge-1.7.10

58 files changed, 3075 insertions, 960 deletions

diff --git a/spark-common/src/main/java/me/lucko/spark/common/SparkPlatform.java b/spark-common/src/main/java/me/lucko/spark/common/SparkPlatform.java
index f92abf3..dae04ff 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/SparkPlatform.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/SparkPlatform.java
@@ -45,13 +45,16 @@ import me.lucko.spark.common.monitor.ping.PingStatistics;
 import me.lucko.spark.common.monitor.ping.PlayerPingProvider;
 import me.lucko.spark.common.monitor.tick.TickStatistics;
 import me.lucko.spark.common.platform.PlatformStatisticsProvider;
+import me.lucko.spark.common.sampler.BackgroundSamplerManager;
+import me.lucko.spark.common.sampler.SamplerContainer;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
 import me.lucko.spark.common.tick.TickHook;
 import me.lucko.spark.common.tick.TickReporter;
 import me.lucko.spark.common.util.BytebinClient;
-import me.lucko.spark.common.util.ClassSourceLookup;
 import me.lucko.spark.common.util.Configuration;
 import me.lucko.spark.common.util.TemporaryFiles;
 
+import net.kyori.adventure.text.Component;
 import net.kyori.adventure.text.event.ClickEvent;
 
 import java.io.IOException;
@@ -62,6 +65,7 @@ import java.time.format.DateTimeFormatter;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Collections;
+import java.util.LinkedHashMap;
 import java.util.List;
 import java.util.Map;
 import java.util.concurrent.atomic.AtomicBoolean;
@@ -72,13 +76,11 @@ import java.util.stream.Collectors;
 
 import static net.kyori.adventure.text.Component.space;
 import static net.kyori.adventure.text.Component.text;
-import static net.kyori.adventure.text.format.NamedTextColor.DARK_GRAY;
 import static net.kyori.adventure.text.format.NamedTextColor.GOLD;
 import static net.kyori.adventure.text.format.NamedTextColor.GRAY;
 import static net.kyori.adventure.text.format.NamedTextColor.RED;
 import static net.kyori.adventure.text.format.NamedTextColor.WHITE;
 import static net.kyori.adventure.text.format.TextDecoration.BOLD;
-import static net.kyori.adventure.text.format.TextDecoration.UNDERLINED;
 
 /**
  * Abstract spark implementation used by all platforms.
@@ -89,6 +91,7 @@ public class SparkPlatform {
     private static final DateTimeFormatter DATE_TIME_FORMATTER = DateTimeFormatter.ofPattern("yyyy-MM-dd_HH.mm.ss");
 
     private final SparkPlugin plugin;
+    private final TemporaryFiles temporaryFiles;
     private final Configuration configuration;
     private final String viewerUrl;
     private final BytebinClient bytebinClient;
@@ -97,6 +100,8 @@ public class SparkPlatform {
     private final List<Command> commands;
     private final ReentrantLock commandExecuteLock = new ReentrantLock(true);
     private final ActivityLog activityLog;
+    private final SamplerContainer samplerContainer;
+    private final BackgroundSamplerManager backgroundSamplerManager;
     private final TickHook tickHook;
     private final TickReporter tickReporter;
     private final TickStatistics tickStatistics;
@@ -109,6 +114,7 @@ public class SparkPlatform {
     public SparkPlatform(SparkPlugin plugin) {
         this.plugin = plugin;
 
+        this.temporaryFiles = new TemporaryFiles(this.plugin.getPluginDirectory().resolve("tmp"));
         this.configuration = new Configuration(this.plugin.getPluginDirectory().resolve("config.json"));
 
         this.viewerUrl = this.configuration.getString("viewerUrl", "https://spark.lucko.me/");
@@ -135,6 +141,9 @@ public class SparkPlatform {
         this.activityLog = new ActivityLog(plugin.getPluginDirectory().resolve("activity.json"));
         this.activityLog.load();
 
+        this.samplerContainer = new SamplerContainer();
+        this.backgroundSamplerManager = new BackgroundSamplerManager(this, this.configuration);
+
         this.tickHook = plugin.createTickHook();
         this.tickReporter = plugin.createTickReporter();
         this.tickStatistics = this.tickHook != null || this.tickReporter != null ? new TickStatistics() : null;
@@ -173,6 +182,8 @@ public class SparkPlatform {
         SparkApi api = new SparkApi(this);
         this.plugin.registerApi(api);
         SparkApi.register(api);
+
+        this.backgroundSamplerManager.initialise();
     }
 
     public void disable() {
@@ -190,15 +201,21 @@ public class SparkPlatform {
             module.close();
         }
 
+        this.samplerContainer.close();
+
         SparkApi.unregister();
 
-        TemporaryFiles.deleteTemporaryFiles();
+        this.temporaryFiles.deleteTemporaryFiles();
     }
 
     public SparkPlugin getPlugin() {
         return this.plugin;
     }
 
+    public TemporaryFiles getTemporaryFiles() {
+        return this.temporaryFiles;
+    }
+
     public Configuration getConfiguration() {
         return this.configuration;
     }
@@ -223,6 +240,14 @@ public class SparkPlatform {
         return this.activityLog;
     }
 
+    public SamplerContainer getSamplerContainer() {
+        return this.samplerContainer;
+    }
+
+    public BackgroundSamplerManager getBackgroundSamplerManager() {
+        return this.backgroundSamplerManager;
+    }
+
     public TickHook getTickHook() {
         return this.tickHook;
     }
@@ -356,14 +381,15 @@ public class SparkPlatform {
                     .append(text("v" + getPlugin().getVersion(), GRAY))
                     .build()
             );
+
+            String helpCmd = "/" + getPlugin().getCommandName() + " help";
             resp.replyPrefixed(text()
                     .color(GRAY)
-                    .append(text("Use "))
+                    .append(text("Run "))
                     .append(text()
-                            .content("/" + getPlugin().getCommandName() + " help")
+                            .content(helpCmd)
                             .color(WHITE)
-                            .decoration(UNDERLINED, true)
-                            .clickEvent(ClickEvent.runCommand("/" + getPlugin().getCommandName() + " help"))
+                            .clickEvent(ClickEvent.runCommand(helpCmd))
                             .build()
                     )
                     .append(text(" to view usage information."))
@@ -379,7 +405,7 @@ public class SparkPlatform {
             if (command.aliases().contains(alias)) {
                 resp.setCommandPrimaryAlias(command.primaryAlias());
                 try {
-                    command.executor().execute(this, sender, resp, new Arguments(rawArgs));
+                    command.executor().execute(this, sender, resp, new Arguments(rawArgs, command.allowSubCommand()));
                 } catch (Arguments.ParseException e) {
                     resp.replyPrefixed(text(e.getMessage(), RED));
                 }
@@ -427,35 +453,53 @@ public class SparkPlatform {
         );
         for (Command command : commands) {
             String usage = "/" + getPlugin().getCommandName() + " " + command.primaryAlias();
-            ClickEvent clickEvent = ClickEvent.suggestCommand(usage);
-            sender.reply(text()
-                    .append(text(">", GOLD, BOLD))
-                    .append(space())
-                    .append(text().content(usage).color(GRAY).clickEvent(clickEvent).build())
-                    .build()
-            );
-            for (Command.ArgumentInfo arg : command.arguments()) {
-                if (arg.requiresParameter()) {
+
+            if (command.allowSubCommand()) {
+                Map<String, List<Command.ArgumentInfo>> argumentsBySubCommand = command.arguments().stream()
+                        .collect(Collectors.groupingBy(Command.ArgumentInfo::subCommandName, LinkedHashMap::new, Collectors.toList()));
+
+                argumentsBySubCommand.forEach((subCommand, arguments) -> {
+                    String subCommandUsage = usage + " " + subCommand;
+
                     sender.reply(text()
-                            .content("       ")
-                            .append(text("[", DARK_GRAY))
-                            .append(text("--" + arg.argumentName(), GRAY))
+                            .append(text(">", GOLD, BOLD))
                             .append(space())
-                            .append(text("<" + arg.parameterDescription() + ">", DARK_GRAY))
-                            .append(text("]", DARK_GRAY))
-                            .build()
-                    );
-                } else {
-                    sender.reply(text()
-                            .content("       ")
-                            .append(text("[", DARK_GRAY))
-                            .append(text("--" + arg.argumentName(), GRAY))
-                            .append(text("]", DARK_GRAY))
+                            .append(text().content(subCommandUsage).color(GRAY).clickEvent(ClickEvent.suggestCommand(subCommandUsage)).build())
                             .build()
                     );
+
+                    for (Command.ArgumentInfo arg : arguments) {
+                        if (arg.argumentName().isEmpty()) {
+                            continue;
+                        }
+                        sender.reply(arg.toComponent("      "));
+                    }
+                });
+            } else {
+                sender.reply(text()
+                        .append(text(">", GOLD, BOLD))
+                        .append(space())
+                        .append(text().content(usage).color(GRAY).clickEvent(ClickEvent.suggestCommand(usage)).build())
+                        .build()
+                );
+
+                for (Command.ArgumentInfo arg : command.arguments()) {
+                    sender.reply(arg.toComponent("    "));
                 }
             }
         }
+
+        sender.reply(Component.empty());
+        sender.replyPrefixed(text()
+                .append(text("For full usage information, please go to: "))
+                .append(text()
+                        .content("https://spark.lucko.me/docs/Command-Usage")
+                        .color(WHITE)
+                        .clickEvent(ClickEvent.openUrl("https://spark.lucko.me/docs/Command-Usage"))
+                        .build()
+                )
+                .build()
+        );
     }
 
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/SparkPlugin.java b/spark-common/src/main/java/me/lucko/spark/common/SparkPlugin.java
index 1116b04..b7aef2a 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/SparkPlugin.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/SparkPlugin.java
@@ -23,15 +23,19 @@ package me.lucko.spark.common;
 import me.lucko.spark.api.Spark;
 import me.lucko.spark.common.command.sender.CommandSender;
 import me.lucko.spark.common.monitor.ping.PlayerPingProvider;
+import me.lucko.spark.common.platform.MetadataProvider;
 import me.lucko.spark.common.platform.PlatformInfo;
 import me.lucko.spark.common.platform.serverconfig.ServerConfigProvider;
 import me.lucko.spark.common.platform.world.WorldInfoProvider;
 import me.lucko.spark.common.sampler.ThreadDumper;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
+import me.lucko.spark.common.sampler.source.SourceMetadata;
 import me.lucko.spark.common.tick.TickHook;
 import me.lucko.spark.common.tick.TickReporter;
-import me.lucko.spark.common.util.ClassSourceLookup;
 
 import java.nio.file.Path;
+import java.util.Collection;
+import java.util.Collections;
 import java.util.logging.Level;
 import java.util.stream.Stream;
 
@@ -133,6 +137,15 @@ public interface SparkPlugin {
     }
 
     /**
+     * Gets a list of known sources (plugins/mods) on the platform.
+     *
+     * @return a list of sources
+     */
+    default Collection<SourceMetadata> getKnownSources() {
+        return Collections.emptyList();
+    }
+
+    /**
      * Creates a player ping provider function.
      *
      * <p>Returns {@code null} if the platform does not support querying player pings</p>
@@ -149,7 +162,16 @@ public interface SparkPlugin {
      * @return the server config provider function
      */
     default ServerConfigProvider createServerConfigProvider() {
-        return ServerConfigProvider.NO_OP;
+        return null;
+    }
+
+    /**
+     * Creates a metadata provider for the platform.
+     *
+     * @return the platform extra metadata provider
+     */
+    default MetadataProvider createExtraMetadataProvider() {
+        return null;
     }
 
     /**
diff --git a/spark-common/src/main/java/me/lucko/spark/common/command/Arguments.java b/spark-common/src/main/java/me/lucko/spark/common/command/Arguments.java
index 17c49e2..ad8c777 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/command/Arguments.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/command/Arguments.java
@@ -38,8 +38,9 @@ public class Arguments {
 
     private final List<String> rawArgs;
     private final SetMultimap<String, String> parsedArgs;
+    private String parsedSubCommand = null;
 
-    public Arguments(List<String> rawArgs) {
+    public Arguments(List<String> rawArgs, boolean allowSubCommand) {
         this.rawArgs = rawArgs;
         this.parsedArgs = HashMultimap.create();
 
@@ -52,7 +53,9 @@ public class Arguments {
             Matcher matcher = FLAG_REGEX.matcher(arg);
             boolean matches = matcher.matches();
 
-            if (flag == null || matches) {
+            if (i == 0 && allowSubCommand && !matches) {
+                this.parsedSubCommand = arg;
+            } else if (flag == null || matches) {
                 if (!matches) {
                     throw new ParseException("Expected flag at position " + i + " but got '" + arg + "' instead!");
                 }
@@ -80,6 +83,10 @@ public class Arguments {
         return this.rawArgs;
     }
 
+    public String subCommand() {
+        return this.parsedSubCommand;
+    }
+
     public int intFlag(String key) {
         Iterator<String> it = this.parsedArgs.get(key).iterator();
         if (it.hasNext()) {
diff --git a/spark-common/src/main/java/me/lucko/spark/common/command/Command.java b/spark-common/src/main/java/me/lucko/spark/common/command/Command.java
index dad15e6..c6871a9 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/command/Command.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/command/Command.java
@@ -25,10 +25,17 @@ import com.google.common.collect.ImmutableList;
 import me.lucko.spark.common.SparkPlatform;
 import me.lucko.spark.common.command.sender.CommandSender;
 
+import net.kyori.adventure.text.Component;
+
 import java.util.Collections;
 import java.util.List;
 import java.util.Objects;
 
+import static net.kyori.adventure.text.Component.space;
+import static net.kyori.adventure.text.Component.text;
+import static net.kyori.adventure.text.format.NamedTextColor.DARK_GRAY;
+import static net.kyori.adventure.text.format.NamedTextColor.GRAY;
+
 public class Command {
 
     public static Builder builder() {
@@ -39,12 +46,14 @@ public class Command {
     private final List<ArgumentInfo> arguments;
     private final Executor executor;
     private final TabCompleter tabCompleter;
+    private final boolean allowSubCommand;
 
-    private Command(List<String> aliases, List<ArgumentInfo> arguments, Executor executor, TabCompleter tabCompleter) {
+    private Command(List<String> aliases, List<ArgumentInfo> arguments, Executor executor, TabCompleter tabCompleter, boolean allowSubCommand) {
         this.aliases = aliases;
         this.arguments = arguments;
         this.executor = executor;
         this.tabCompleter = tabCompleter;
+        this.allowSubCommand = allowSubCommand;
     }
 
     public List<String> aliases() {
@@ -67,11 +76,16 @@ public class Command {
         return this.aliases.get(0);
     }
 
+    public boolean allowSubCommand() {
+        return this.allowSubCommand;
+    }
+
     public static final class Builder {
         private final ImmutableList.Builder<String> aliases = ImmutableList.builder();
         private final ImmutableList.Builder<ArgumentInfo> arguments = ImmutableList.builder();
         private Executor executor = null;
         private TabCompleter tabCompleter = null;
+        private boolean allowSubCommand = false;
 
         Builder() {
 
@@ -82,8 +96,13 @@ public class Command {
             return this;
         }
 
+        public Builder argumentUsage(String subCommandName, String argumentName, String parameterDescription) {
+            this.arguments.add(new ArgumentInfo(subCommandName, argumentName, parameterDescription));
+            return this;
+        }
+
         public Builder argumentUsage(String argumentName, String parameterDescription) {
-            this.arguments.add(new ArgumentInfo(argumentName, parameterDescription));
+            this.arguments.add(new ArgumentInfo("", argumentName, parameterDescription));
             return this;
         }
 
@@ -97,6 +116,11 @@ public class Command {
             return this;
         }
 
+        public Builder allowSubCommand(boolean allowSubCommand) {
+            this.allowSubCommand = allowSubCommand;
+            return this;
+        }
+
         public Command build() {
             List<String> aliases = this.aliases.build();
             if (aliases.isEmpty()) {
@@ -108,7 +132,7 @@ public class Command {
             if (this.tabCompleter == null) {
                 this.tabCompleter = TabCompleter.empty();
             }
-            return new Command(aliases, this.arguments.build(), this.executor, this.tabCompleter);
+            return new Command(aliases, this.arguments.build(), this.executor, this.tabCompleter, this.allowSubCommand);
         }
     }
 
@@ -127,14 +151,20 @@ public class Command {
     }
 
     public static final class ArgumentInfo {
+        private final String subCommandName;
         private final String argumentName;
         private final String parameterDescription;
 
-        public ArgumentInfo(String argumentName, String parameterDescription) {
+        public ArgumentInfo(String subCommandName, String argumentName, String parameterDescription) {
+            this.subCommandName = subCommandName;
             this.argumentName = argumentName;
             this.parameterDescription = parameterDescription;
         }
 
+        public String subCommandName() {
+            return this.subCommandName;
+        }
+
         public String argumentName() {
             return this.argumentName;
         }
@@ -146,6 +176,26 @@ public class Command {
         public boolean requiresParameter() {
             return this.parameterDescription != null;
         }
+
+        public Component toComponent(String padding) {
+            if (requiresParameter()) {
+                 return text()
+                        .content(padding)
+                        .append(text("[", DARK_GRAY))
+                        .append(text("--" + argumentName(), GRAY))
+                        .append(space())
+                        .append(text("<" + parameterDescription() + ">", DARK_GRAY))
+                        .append(text("]", DARK_GRAY))
+                        .build();
+            } else {
+                return text()
+                        .content(padding)
+                        .append(text("[", DARK_GRAY))
+                        .append(text("--" + argumentName(), GRAY))
+                        .append(text("]", DARK_GRAY))
+                        .build();
+            }
+        }
     }
 
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/command/modules/GcMonitoringModule.java b/spark-common/src/main/java/me/lucko/spark/common/command/modules/GcMonitoringModule.java
index 2ce83fd..a2da0a0 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/command/modules/GcMonitoringModule.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/command/modules/GcMonitoringModule.java
@@ -123,7 +123,7 @@ public class GcMonitoringModule implements CommandModule {
                         );
                         report.add(text()
                                 .content("      ")
-                                .append(text(formatTime((long) averageFrequency), WHITE))
+                                .append(text(FormatUtil.formatSeconds((long) averageFrequency / 1000), WHITE))
                                 .append(text(" avg frequency", GRAY))
                                 .build()
                         );
@@ -153,26 +153,6 @@ public class GcMonitoringModule implements CommandModule {
         );
     }
 
-    private static String formatTime(long millis) {
-        if (millis <= 0) {
-            return "0s";
-        }
-
-        long second = millis / 1000;
-        long minute = second / 60;
-        second = second % 60;
-
-        StringBuilder sb = new StringBuilder();
-        if (minute != 0) {
-            sb.append(minute).append("m ");
-        }
-        if (second != 0) {
-            sb.append(second).append("s ");
-        }
-
-        return sb.toString().trim();
-    }
-
     private static class ReportingGcMonitor extends GarbageCollectionMonitor implements GarbageCollectionMonitor.Listener {
         private final SparkPlatform platform;
         private final CommandResponseHandler resp;
diff --git a/spark-common/src/main/java/me/lucko/spark/common/command/modules/SamplerModule.java b/spark-common/src/main/java/me/lucko/spark/common/command/modules/SamplerModule.java
index 0a80c31..cd00f0d 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/command/modules/SamplerModule.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/command/modules/SamplerModule.java
@@ -35,13 +35,15 @@ import me.lucko.spark.common.sampler.Sampler;
 import me.lucko.spark.common.sampler.SamplerBuilder;
 import me.lucko.spark.common.sampler.ThreadDumper;
 import me.lucko.spark.common.sampler.ThreadGrouper;
-import me.lucko.spark.common.sampler.ThreadNodeOrder;
 import me.lucko.spark.common.sampler.async.AsyncSampler;
 import me.lucko.spark.common.sampler.node.MergeMode;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
 import me.lucko.spark.common.tick.TickHook;
+import me.lucko.spark.common.util.FormatUtil;
 import me.lucko.spark.common.util.MethodDisambiguator;
 import me.lucko.spark.proto.SparkSamplerProtos;
 
+import net.kyori.adventure.text.Component;
 import net.kyori.adventure.text.event.ClickEvent;
 
 import java.io.IOException;
@@ -62,57 +64,45 @@ import static net.kyori.adventure.text.format.NamedTextColor.DARK_GRAY;
 import static net.kyori.adventure.text.format.NamedTextColor.GOLD;
 import static net.kyori.adventure.text.format.NamedTextColor.GRAY;
 import static net.kyori.adventure.text.format.NamedTextColor.RED;
+import static net.kyori.adventure.text.format.NamedTextColor.WHITE;
 
 public class SamplerModule implements CommandModule {
     private static final String SPARK_SAMPLER_MEDIA_TYPE = "application/x-spark-sampler";
 
-    /** The sampler instance currently running, if any */
-    private Sampler activeSampler = null;
-
-    @Override
-    public void close() {
-        if (this.activeSampler != null) {
-            this.activeSampler.stop();
-            this.activeSampler = null;
-        }
-    }
-
     @Override
     public void registerCommands(Consumer<Command> consumer) {
         consumer.accept(Command.builder()
                 .aliases("profiler", "sampler")
-                .argumentUsage("info", null)
-                .argumentUsage("stop", null)
-                .argumentUsage("cancel", null)
-                .argumentUsage("interval", "interval millis")
-                .argumentUsage("thread", "thread name")
-                .argumentUsage("only-ticks-over", "tick length millis")
-                .argumentUsage("timeout", "timeout seconds")
-                .argumentUsage("regex --thread", "thread regex")
-                .argumentUsage("combine-all", null)
-                .argumentUsage("not-combined", null)
-                .argumentUsage("force-java-sampler", null)
-                .argumentUsage("stop --comment", "comment")
-                .argumentUsage("stop --order-by-time", null)
-                .argumentUsage("stop --save-to-file", null)
+                .allowSubCommand(true)
+                .argumentUsage("info", "", null)
+                .argumentUsage("start", "timeout", "timeout seconds")
+                .argumentUsage("start", "thread *", null)
+                .argumentUsage("start", "thread", "thread name")
+                .argumentUsage("start", "only-ticks-over", "tick length millis")
+                .argumentUsage("start", "interval", "interval millis")
+                .argumentUsage("stop", "", null)
+                .argumentUsage("cancel", "", null)
                 .executor(this::profiler)
                 .tabCompleter((platform, sender, arguments) -> {
-                    if (arguments.contains("--info") || arguments.contains("--cancel")) {
-                        return Collections.emptyList();
-                    }
-
-                    if (arguments.contains("--stop") || arguments.contains("--upload")) {
-                        return TabCompleter.completeForOpts(arguments, "--order-by-time", "--comment", "--save-to-file");
+                    List<String> opts = Collections.emptyList();
+
+                    if (arguments.size() > 0) {
+                        String subCommand = arguments.get(0);
+                        if (subCommand.equals("stop") || subCommand.equals("upload")) {
+                            opts = new ArrayList<>(Arrays.asList("--comment", "--save-to-file"));
+                            opts.removeAll(arguments);
+                        }
+                        if (subCommand.equals("start")) {
+                            opts = new ArrayList<>(Arrays.asList("--timeout", "--regex", "--combine-all",
+                                    "--not-combined", "--interval", "--only-ticks-over", "--force-java-sampler"));
+                            opts.removeAll(arguments);
+                            opts.add("--thread"); // allowed multiple times
+                        }
                     }
 
-                    List<String> opts = new ArrayList<>(Arrays.asList("--info", "--stop", "--cancel",
-                            "--timeout", "--regex", "--combine-all", "--not-combined", "--interval",
-                            "--only-ticks-over", "--force-java-sampler"));
-                    opts.removeAll(arguments);
-                    opts.add("--thread"); // allowed multiple times
-
                     return TabCompleter.create()
-                            .from(0, CompletionSupplier.startsWith(opts))
+                            .at(0, CompletionSupplier.startsWith(Arrays.asList("info", "start", "stop", "cancel")))
+                            .from(1, CompletionSupplier.startsWith(opts))
                             .complete(arguments);
                 })
                 .build()
@@ -120,25 +110,50 @@ public class SamplerModule implements CommandModule {
     }
 
     private void profiler(SparkPlatform platform, CommandSender sender, CommandResponseHandler resp, Arguments arguments) {
-        if (arguments.boolFlag("info")) {
-            profilerInfo(resp);
+        String subCommand = arguments.subCommand() == null ? "" : arguments.subCommand();
+
+        if (subCommand.equals("info") || arguments.boolFlag("info")) {
+            profilerInfo(platform, resp);
             return;
         }
 
-        if (arguments.boolFlag("cancel")) {
-            profilerCancel(resp);
+        if (subCommand.equals("cancel") || arguments.boolFlag("cancel")) {
+            profilerCancel(platform, resp);
             return;
         }
 
-        if (arguments.boolFlag("stop") || arguments.boolFlag("upload")) {
+        if (subCommand.equals("stop") || subCommand.equals("upload") || arguments.boolFlag("stop") || arguments.boolFlag("upload")) {
             profilerStop(platform, sender, resp, arguments);
             return;
         }
 
-        profilerStart(platform, sender, resp, arguments);
+        if (subCommand.equals("start") || arguments.boolFlag("start")) {
+            profilerStart(platform, sender, resp, arguments);
+            return;
+        }
+
+        if (arguments.raw().isEmpty()) {
+            profilerInfo(platform, resp);
+        } else {
+            profilerStart(platform, sender, resp, arguments);
+        }
     }
 
     private void profilerStart(SparkPlatform platform, CommandSender sender, CommandResponseHandler resp, Arguments arguments) {
+        Sampler previousSampler = platform.getSamplerContainer().getActiveSampler();
+        if (previousSampler != null) {
+            if (previousSampler.isRunningInBackground()) {
+                // there is a background profiler running - stop that first
+                resp.replyPrefixed(text("Stopping the background profiler before starting... please wait"));
+                previousSampler.stop(true);
+                platform.getSamplerContainer().unsetActiveSampler(previousSampler);
+            } else {
+                // there is a non-background profiler running - tell the user
+                profilerInfo(platform, resp);
+                return;
+            }
+        }
+
         int timeoutSeconds = arguments.intFlag("timeout");
         if (timeoutSeconds != -1 && timeoutSeconds <= 10) {
             resp.replyPrefixed(text("The specified timeout is not long enough for accurate results to be formed. " +
@@ -195,12 +210,7 @@ public class SamplerModule implements CommandModule {
             }
         }
 
-        if (this.activeSampler != null) {
-            resp.replyPrefixed(text("An active profiler is already running."));
-            return;
-        }
-
-        resp.broadcastPrefixed(text("Initializing a new profiler, please wait..."));
+        resp.broadcastPrefixed(text("Starting a new profiler, please wait..."));
 
         SamplerBuilder builder = new SamplerBuilder();
         builder.threadDumper(threadDumper);
@@ -215,21 +225,25 @@ public class SamplerModule implements CommandModule {
         if (ticksOver != -1) {
             builder.ticksOver(ticksOver, tickHook);
         }
-        Sampler sampler = this.activeSampler = builder.start(platform);
+        Sampler sampler = builder.start(platform);
+        platform.getSamplerContainer().setActiveSampler(sampler);
 
         resp.broadcastPrefixed(text()
-                .append(text("Profiler now active!", GOLD))
+                .append(text("Profiler is now running!", GOLD))
                 .append(space())
                 .append(text("(" + (sampler instanceof AsyncSampler ? "async" : "built-in java") + ")", DARK_GRAY))
                 .build()
         );
+
         if (timeoutSeconds == -1) {
-            resp.broadcastPrefixed(text("Use '/" + platform.getPlugin().getCommandName() + " profiler --stop' to stop profiling and upload the results."));
+            resp.broadcastPrefixed(text("It will run in the background until it is stopped by an admin."));
+            resp.broadcastPrefixed(text("To stop the profiler and upload the results, run:"));
+            resp.broadcastPrefixed(cmdPrompt("/" + platform.getPlugin().getCommandName() + " profiler stop"));
         } else {
-            resp.broadcastPrefixed(text("The results will be automatically returned after the profiler has been running for " + timeoutSeconds + " seconds."));
+            resp.broadcastPrefixed(text("The results will be automatically returned after the profiler has been running for " + FormatUtil.formatSeconds(timeoutSeconds) + "."));
         }
 
-        CompletableFuture<Sampler> future = this.activeSampler.getFuture();
+        CompletableFuture<Sampler> future = sampler.getFuture();
 
         // send message if profiling fails
         future.whenCompleteAsync((s, throwable) -> {
@@ -240,70 +254,101 @@ public class SamplerModule implements CommandModule {
         });
 
         // set activeSampler to null when complete.
-        future.whenCompleteAsync((s, throwable) -> {
-            if (sampler == this.activeSampler) {
-                this.activeSampler = null;
-            }
-        });
+        sampler.getFuture().whenCompleteAsync((s, throwable) -> platform.getSamplerContainer().unsetActiveSampler(s));
 
         // await the result
         if (timeoutSeconds != -1) {
-            ThreadNodeOrder threadOrder = arguments.boolFlag("order-by-time") ? ThreadNodeOrder.BY_TIME : ThreadNodeOrder.BY_NAME;
             String comment = Iterables.getFirst(arguments.stringFlag("comment"), null);
             MethodDisambiguator methodDisambiguator = new MethodDisambiguator();
             MergeMode mergeMode = arguments.boolFlag("separate-parent-calls") ? MergeMode.separateParentCalls(methodDisambiguator) : MergeMode.sameMethod(methodDisambiguator);
             boolean saveToFile = arguments.boolFlag("save-to-file");
             future.thenAcceptAsync(s -> {
                 resp.broadcastPrefixed(text("The active profiler has completed! Uploading results..."));
-                handleUpload(platform, resp, s, threadOrder, comment, mergeMode, saveToFile);
+                handleUpload(platform, resp, s, comment, mergeMode, saveToFile);
             });
         }
     }
 
-    private void profilerInfo(CommandResponseHandler resp) {
-        if (this.activeSampler == null) {
-            resp.replyPrefixed(text("There isn't an active profiler running."));
+    private void profilerInfo(SparkPlatform platform, CommandResponseHandler resp) {
+        Sampler sampler = platform.getSamplerContainer().getActiveSampler();
+        if (sampler == null) {
+            resp.replyPrefixed(text("The profiler isn't running!"));
+            resp.replyPrefixed(text("To start a new one, run:"));
+            resp.replyPrefixed(cmdPrompt("/" + platform.getPlugin().getCommandName() + " profiler start"));
         } else {
-            long timeout = this.activeSampler.getAutoEndTime();
+            resp.replyPrefixed(text("Profiler is already running!", GOLD));
+
+            long runningTime = (System.currentTimeMillis() - sampler.getStartTime()) / 1000L;
+
+            if (sampler.isRunningInBackground()) {
+                resp.replyPrefixed(text()
+                        .append(text("It was started "))
+                        .append(text("automatically", WHITE))
+                        .append(text(" when spark enabled and has been running in the background for " + FormatUtil.formatSeconds(runningTime) + "."))
+                        .build()
+                );
+            } else {
+                resp.replyPrefixed(text("So far, it has profiled for " + FormatUtil.formatSeconds(runningTime) + "."));
+            }
+
+            long timeout = sampler.getAutoEndTime();
             if (timeout == -1) {
-                resp.replyPrefixed(text("There is an active profiler currently running, with no defined timeout."));
+                resp.replyPrefixed(text("To stop the profiler and upload the results, run:"));
+                resp.replyPrefixed(cmdPrompt("/" + platform.getPlugin().getCommandName() + " profiler stop"));
             } else {
                 long timeoutDiff = (timeout - System.currentTimeMillis()) / 1000L;
-                resp.replyPrefixed(text("There is an active profiler currently running, due to timeout in " + timeoutDiff + " seconds."));
+                resp.replyPrefixed(text("It is due to complete automatically and upload results in " + FormatUtil.formatSeconds(timeoutDiff) + "."));
             }
 
-            long runningTime = (System.currentTimeMillis() - this.activeSampler.getStartTime()) / 1000L;
-            resp.replyPrefixed(text("It has been profiling for " + runningTime + " seconds so far."));
+            resp.replyPrefixed(text("To cancel the profiler without uploading the results, run:"));
+            resp.replyPrefixed(cmdPrompt("/" + platform.getPlugin().getCommandName() + " profiler cancel"));
         }
     }
 
-    private void profilerCancel(CommandResponseHandler resp) {
-        if (this.activeSampler == null) {
+    private void profilerCancel(SparkPlatform platform, CommandResponseHandler resp) {
+        Sampler sampler = platform.getSamplerContainer().getActiveSampler();
+        if (sampler == null) {
             resp.replyPrefixed(text("There isn't an active profiler running."));
         } else {
-            close();
-            resp.broadcastPrefixed(text("The active profiler has been cancelled.", GOLD));
+            platform.getSamplerContainer().stopActiveSampler(true);
+            resp.broadcastPrefixed(text("Profiler has been cancelled.", GOLD));
         }
     }
 
     private void profilerStop(SparkPlatform platform, CommandSender sender, CommandResponseHandler resp, Arguments arguments) {
-        if (this.activeSampler == null) {
+        Sampler sampler = platform.getSamplerContainer().getActiveSampler();
+
+        if (sampler == null) {
             resp.replyPrefixed(text("There isn't an active profiler running."));
         } else {
-            this.activeSampler.stop();
-            resp.broadcastPrefixed(text("The active profiler has been stopped! Uploading results..."));
-            ThreadNodeOrder threadOrder = arguments.boolFlag("order-by-time") ? ThreadNodeOrder.BY_TIME : ThreadNodeOrder.BY_NAME;
+            platform.getSamplerContainer().unsetActiveSampler(sampler);
+            sampler.stop(false);
+
+            boolean saveToFile = arguments.boolFlag("save-to-file");
+            if (saveToFile) {
+                resp.broadcastPrefixed(text("Stopping the profiler & saving results, please wait..."));
+            } else {
+                resp.broadcastPrefixed(text("Stopping the profiler & uploading results, please wait..."));
+            }
+
             String comment = Iterables.getFirst(arguments.stringFlag("comment"), null);
             MethodDisambiguator methodDisambiguator = new MethodDisambiguator();
             MergeMode mergeMode = arguments.boolFlag("separate-parent-calls") ? MergeMode.separateParentCalls(methodDisambiguator) : MergeMode.sameMethod(methodDisambiguator);
-            boolean saveToFile = arguments.boolFlag("save-to-file");
-            handleUpload(platform, resp, this.activeSampler, threadOrder, comment, mergeMode, saveToFile);
-            this.activeSampler = null;
+            handleUpload(platform, resp, sampler, comment, mergeMode, saveToFile);
+
+            // if the previous sampler was running in the background, create a new one
+            if (platform.getBackgroundSamplerManager().restartBackgroundSampler()) {
+                resp.broadcastPrefixed(text()
+                        .append(text("Restarted the background profiler. "))
+                        .append(text("(If you don't want this to happen, run: /" + platform.getPlugin().getCommandName() + " profiler cancel)", DARK_GRAY))
+                        .build()
+                );
+            }
         }
     }
 
-    private void handleUpload(SparkPlatform platform, CommandResponseHandler resp, Sampler sampler, ThreadNodeOrder threadOrder, String comment, MergeMode mergeMode, boolean saveToFileFlag) {
-        SparkSamplerProtos.SamplerData output = sampler.toProto(platform, resp.sender(), threadOrder, comment, mergeMode, platform.createClassSourceLookup());
+    private void handleUpload(SparkPlatform platform, CommandResponseHandler resp, Sampler sampler, String comment, MergeMode mergeMode, boolean saveToFileFlag) {
+        SparkSamplerProtos.SamplerData output = sampler.toProto(platform, resp.sender(), comment, mergeMode, ClassSourceLookup.create(platform));
 
         boolean saveToFile = false;
         if (saveToFileFlag) {
@@ -313,7 +358,7 @@ public class SamplerModule implements CommandModule {
                 String key = platform.getBytebinClient().postContent(output, SPARK_SAMPLER_MEDIA_TYPE).key();
                 String url = platform.getViewerUrl() + key;
 
-                resp.broadcastPrefixed(text("Profiler results:", GOLD));
+                resp.broadcastPrefixed(text("Profiler stopped & upload complete!", GOLD));
                 resp.broadcast(text()
                         .content(url)
                         .color(GRAY)
@@ -334,13 +379,9 @@ public class SamplerModule implements CommandModule {
             try {
                 Files.write(file, output.toByteArray());
 
-                resp.broadcastPrefixed(text()
-                        .content("Profile written to: ")
-                        .color(GOLD)
-                        .append(text(file.toString(), GRAY))
-                        .build()
-                );
-                resp.broadcastPrefixed(text("You can read the profile file using the viewer web-app - " + platform.getViewerUrl(), GRAY));
+                resp.broadcastPrefixed(text("Profiler stopped & save complete!", GOLD));
+                resp.broadcastPrefixed(text("Data has been written to: " + file));
+                resp.broadcastPrefixed(text("You can view the profile file using the web app @ " + platform.getViewerUrl(), GRAY));
 
                 platform.getActivityLog().addToLog(Activity.fileActivity(resp.sender(), System.currentTimeMillis(), "Profiler", file.toString()));
             } catch (IOException e) {
@@ -349,4 +390,16 @@ public class SamplerModule implements CommandModule {
             }
         }
     }
+
+    private static Component cmdPrompt(String cmd) {
+        return text()
+                .append(text("  "))
+                .append(text()
+                        .content(cmd)
+                        .color(WHITE)
+                        .clickEvent(ClickEvent.runCommand(cmd))
+                        .build()
+                )
+                .build();
+    }
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/util/LinuxProc.java b/spark-common/src/main/java/me/lucko/spark/common/monitor/LinuxProc.java
index 7d688d7..563e247 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/util/LinuxProc.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/monitor/LinuxProc.java
@@ -18,7 +18,7 @@
  *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-package me.lucko.spark.common.util;
+package me.lucko.spark.common.monitor;
 
 import org.checkerframework.checker.nullness.qual.NonNull;
 import org.checkerframework.checker.nullness.qual.Nullable;
@@ -49,7 +49,12 @@ public enum LinuxProc {
     /**
      * Information about the system network usage.
      */
-    NET_DEV("/proc/net/dev");
+    NET_DEV("/proc/net/dev"),
+
+    /**
+     * Information about the operating system distro.
+     */
+    OSINFO("/etc/os-release");
 
     private final Path path;
 
diff --git a/spark-common/src/main/java/me/lucko/spark/common/monitor/WindowsWmic.java b/spark-common/src/main/java/me/lucko/spark/common/monitor/WindowsWmic.java
new file mode 100644
index 0000000..6b602d9
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/monitor/WindowsWmic.java
@@ -0,0 +1,74 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.monitor;
+
+import org.checkerframework.checker.nullness.qual.NonNull;
+
+import java.io.BufferedReader;
+import java.io.InputStreamReader;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.List;
+
+/**
+ * Utility for reading from wmic (Windows Management Instrumentation Commandline) on Windows systems.
+ */
+public enum WindowsWmic {
+
+    /**
+     * Gets the CPU name
+     */
+    CPU_GET_NAME("wmic", "cpu", "get", "name", "/FORMAT:list"),
+
+    /**
+     * Gets the operating system name (caption) and version.
+     */
+    OS_GET_CAPTION_AND_VERSION("wmic", "os", "get", "caption,version", "/FORMAT:list");
+
+    private static final boolean SUPPORTED = System.getProperty("os.name").startsWith("Windows");
+
+    private final String[] cmdArgs;
+
+    WindowsWmic(String... cmdArgs) {
+        this.cmdArgs = cmdArgs;
+    }
+
+    public @NonNull List<String> read() {
+        if (SUPPORTED) {
+            ProcessBuilder process = new ProcessBuilder(this.cmdArgs).redirectErrorStream(true);
+            try (BufferedReader buf = new BufferedReader(new InputStreamReader(process.start().getInputStream()))) {
+                List<String> lines = new ArrayList<>();
+
+                String line;
+                while ((line = buf.readLine()) != null) {
+                    lines.add(line);
+                }
+
+                return lines;
+            } catch (Exception e) {
+                // ignore
+            }
+        }
+
+        return Collections.emptyList();
+    }
+}
+
diff --git a/spark-common/src/main/java/me/lucko/spark/common/monitor/cpu/CpuInfo.java b/spark-common/src/main/java/me/lucko/spark/common/monitor/cpu/CpuInfo.java
index 9bbe0f8..9954bd5 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/monitor/cpu/CpuInfo.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/monitor/cpu/CpuInfo.java
@@ -20,12 +20,13 @@
 
 package me.lucko.spark.common.monitor.cpu;
 
-import me.lucko.spark.common.util.LinuxProc;
+import me.lucko.spark.common.monitor.LinuxProc;
+import me.lucko.spark.common.monitor.WindowsWmic;
 
 import java.util.regex.Pattern;
 
 /**
- * Small utility to query the CPU model on Linux systems.
+ * Small utility to query the CPU model on Linux and Windows systems.
  */
 public enum CpuInfo {
     ;
@@ -40,11 +41,17 @@ public enum CpuInfo {
     public static String queryCpuModel() {
         for (String line : LinuxProc.CPUINFO.read()) {
             String[] splitLine = SPACE_COLON_SPACE_PATTERN.split(line);
-
             if (splitLine[0].equals("model name") || splitLine[0].equals("Processor")) {
                 return splitLine[1];
             }
         }
+
+        for (String line : WindowsWmic.CPU_GET_NAME.read()) {
+            if (line.startsWith("Name")) {
+                return line.substring(5).trim();
+            }
+        }
+
         return "";
     }
 
diff --git a/spark-common/src/main/java/me/lucko/spark/common/monitor/memory/MemoryInfo.java b/spark-common/src/main/java/me/lucko/spark/common/monitor/memory/MemoryInfo.java
index 226f75b..8f63f71 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/monitor/memory/MemoryInfo.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/monitor/memory/MemoryInfo.java
@@ -20,7 +20,7 @@
 
 package me.lucko.spark.common.monitor.memory;
 
-import me.lucko.spark.common.util.LinuxProc;
+import me.lucko.spark.common.monitor.LinuxProc;
 
 import java.lang.management.ManagementFactory;
 import java.util.regex.Matcher;
diff --git a/spark-common/src/main/java/me/lucko/spark/common/monitor/net/NetworkInterfaceInfo.java b/spark-common/src/main/java/me/lucko/spark/common/monitor/net/NetworkInterfaceInfo.java
index bd9e187..332077a 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/monitor/net/NetworkInterfaceInfo.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/monitor/net/NetworkInterfaceInfo.java
@@ -22,7 +22,7 @@ package me.lucko.spark.common.monitor.net;
 
 import com.google.common.collect.ImmutableMap;
 
-import me.lucko.spark.common.util.LinuxProc;
+import me.lucko.spark.common.monitor.LinuxProc;
 
 import org.checkerframework.checker.nullness.qual.NonNull;
 
diff --git a/spark-common/src/main/java/me/lucko/spark/common/monitor/os/OperatingSystemInfo.java b/spark-common/src/main/java/me/lucko/spark/common/monitor/os/OperatingSystemInfo.java
new file mode 100644
index 0000000..1c2732c
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/monitor/os/OperatingSystemInfo.java
@@ -0,0 +1,86 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.monitor.os;
+
+import me.lucko.spark.common.monitor.LinuxProc;
+import me.lucko.spark.common.monitor.WindowsWmic;
+
+/**
+ * Small utility to query the operating system name & version.
+ */
+public final class OperatingSystemInfo {
+    private final String name;
+    private final String version;
+    private final String arch;
+
+    public OperatingSystemInfo(String name, String version, String arch) {
+        this.name = name;
+        this.version = version;
+        this.arch = arch;
+    }
+
+    public String name() {
+        return this.name;
+    }
+
+    public String version() {
+        return this.version;
+    }
+
+    public String arch() {
+        return this.arch;
+    }
+
+    public static OperatingSystemInfo poll() {
+        String name = null;
+        String version = null;
+
+        for (String line : LinuxProc.OSINFO.read()) {
+            if (line.startsWith("PRETTY_NAME") && line.length() > 13) {
+                name = line.substring(13).replace('"', ' ').trim();
+            }
+        }
+
+        for (String line : WindowsWmic.OS_GET_CAPTION_AND_VERSION.read()) {
+            if (line.startsWith("Caption") && line.length() > 18) {
+                // Caption=Microsoft Windows something
+                // \----------------/ = 18 chars
+                name = line.substring(18).trim();
+            } else if (line.startsWith("Version")) {
+                // Version=10.0.something
+                // \------/ = 8 chars
+                version = line.substring(8).trim();
+            }
+        }
+
+        if (name == null) {
+            name = System.getProperty("os.name");
+        }
+
+        if (version == null) {
+            version = System.getProperty("os.version");
+        }
+
+        String arch = System.getProperty("os.arch");
+
+        return new OperatingSystemInfo(name, version, arch);
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/ThreadNodeOrder.java b/spark-common/src/main/java/me/lucko/spark/common/platform/MetadataProvider.java
index adcedcd..39022b4 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/ThreadNodeOrder.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/platform/MetadataProvider.java
@@ -18,35 +18,30 @@
  *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-package me.lucko.spark.common.sampler;
+package me.lucko.spark.common.platform;
 
-import me.lucko.spark.common.sampler.node.ThreadNode;
+import com.google.gson.JsonElement;
 
-import java.util.Comparator;
+import java.util.LinkedHashMap;
+import java.util.Map;
 
 /**
- * Methods of ordering {@link ThreadNode}s in the output data.
+ * Function to export dynamic metadata to be displayed within the spark viewer.
  */
-public enum ThreadNodeOrder implements Comparator<ThreadNode> {
+@FunctionalInterface
+public interface MetadataProvider {
 
     /**
-     * Order by the name of the thread (alphabetically)
+     * Produces a map of the metadata.
+     *
+     * @return the metadata
      */
-    BY_NAME {
-        @Override
-        public int compare(ThreadNode o1, ThreadNode o2) {
-            return o1.getThreadLabel().compareTo(o2.getThreadLabel());
-        }
-    },
+    Map<String, JsonElement> get();
 
-    /**
-     * Order by the time taken by the thread (most time taken first)
-     */
-    BY_TIME {
-        @Override
-        public int compare(ThreadNode o1, ThreadNode o2) {
-            return -Double.compare(o1.getTotalTime(), o2.getTotalTime());
-        }
+    default Map<String, String> export() {
+        Map<String, String> map = new LinkedHashMap<>();
+        get().forEach((key, value) -> map.put(key, value.toString()));
+        return map;
     }
 
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/platform/PlatformStatisticsProvider.java b/spark-common/src/main/java/me/lucko/spark/common/platform/PlatformStatisticsProvider.java
index 49cfed5..fc7e78a 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/platform/PlatformStatisticsProvider.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/platform/PlatformStatisticsProvider.java
@@ -28,9 +28,10 @@ import me.lucko.spark.common.monitor.memory.GarbageCollectorStatistics;
 import me.lucko.spark.common.monitor.memory.MemoryInfo;
 import me.lucko.spark.common.monitor.net.NetworkInterfaceAverages;
 import me.lucko.spark.common.monitor.net.NetworkMonitor;
+import me.lucko.spark.common.monitor.os.OperatingSystemInfo;
 import me.lucko.spark.common.monitor.ping.PingStatistics;
 import me.lucko.spark.common.monitor.tick.TickStatistics;
-import me.lucko.spark.common.platform.world.WorldInfoProvider;
+import me.lucko.spark.common.platform.world.AsyncWorldInfoProvider;
 import me.lucko.spark.common.platform.world.WorldStatisticsProvider;
 import me.lucko.spark.proto.SparkProtos.PlatformStatistics;
 import me.lucko.spark.proto.SparkProtos.SystemStatistics;
@@ -50,6 +51,7 @@ public class PlatformStatisticsProvider {
 
     public SystemStatistics getSystemStatistics() {
         RuntimeMXBean runtimeBean = ManagementFactory.getRuntimeMXBean();
+        OperatingSystemInfo osInfo = OperatingSystemInfo.poll();
 
         SystemStatistics.Builder builder = SystemStatistics.newBuilder()
                 .setCpu(SystemStatistics.Cpu.newBuilder()
@@ -86,9 +88,9 @@ public class PlatformStatisticsProvider {
                         .build()
                 )
                 .setOs(SystemStatistics.Os.newBuilder()
-                        .setArch(System.getProperty("os.arch"))
-                        .setName(System.getProperty("os.name"))
-                        .setVersion(System.getProperty("os.version"))
+                        .setArch(osInfo.arch())
+                        .setName(osInfo.name())
+                        .setVersion(osInfo.version())
                         .build()
                 )
                 .setJava(SystemStatistics.Java.newBuilder()
@@ -186,8 +188,9 @@ public class PlatformStatisticsProvider {
         }
 
         try {
-            WorldInfoProvider worldInfo = this.platform.getPlugin().createWorldInfoProvider();
-            WorldStatisticsProvider worldStatisticsProvider = new WorldStatisticsProvider(this.platform, worldInfo);
+            WorldStatisticsProvider worldStatisticsProvider = new WorldStatisticsProvider(
+                    new AsyncWorldInfoProvider(this.platform, this.platform.getPlugin().createWorldInfoProvider())
+            );
             WorldStatistics worldStatistics = worldStatisticsProvider.getWorldStatistics();
             if (worldStatistics != null) {
                 builder.setWorld(worldStatistics);
diff --git a/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/AbstractServerConfigProvider.java b/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/AbstractServerConfigProvider.java
deleted file mode 100644
index 559ae95..0000000
--- a/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/AbstractServerConfigProvider.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/*
- * This file is part of spark.
- *
- *  Copyright (c) lucko (Luck) <luck@lucko.me>
- *  Copyright (c) contributors
- *
- *  This program is free software: you can redistribute it and/or modify
- *  it under the terms of the GNU General Public License as published by
- *  the Free Software Foundation, either version 3 of the License, or
- *  (at your option) any later version.
- *
- *  This program is distributed in the hope that it will be useful,
- *  but WITHOUT ANY WARRANTY; without even the implied warranty of
- *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *  GNU General Public License for more details.
- *
- *  You should have received a copy of the GNU General Public License
- *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
- */
-
-package me.lucko.spark.common.platform.serverconfig;
-
-import com.google.common.collect.ImmutableMap;
-import com.google.gson.JsonElement;
-
-import java.util.Arrays;
-import java.util.Collection;
-import java.util.Collections;
-import java.util.List;
-import java.util.Map;
-
-/**
- * Abstract implementation of {@link ServerConfigProvider}.
- *
- * <p>This implementation is able to delete hidden paths from
- * the configurations before they are sent to the viewer.</p>
- */
-public abstract class AbstractServerConfigProvider implements ServerConfigProvider {
-    private final Map<String, ConfigParser> files;
-    private final ExcludedConfigFilter hiddenPathFilters;
-
-    protected AbstractServerConfigProvider(Map<String, ConfigParser> files, Collection<String> hiddenPaths) {
-        this.files = files;
-        this.hiddenPathFilters = new ExcludedConfigFilter(hiddenPaths);
-    }
-
-    @Override
-    public final Map<String, JsonElement> loadServerConfigurations() {
-        ImmutableMap.Builder<String, JsonElement> builder = ImmutableMap.builder();
-
-        this.files.forEach((path, parser) -> {
-            try {
-                JsonElement json = parser.load(path, this.hiddenPathFilters);
-                if (json == null) {
-                    return;
-                }
-                builder.put(path, json);
-            } catch (Exception e) {
-                e.printStackTrace();
-            }
-        });
-
-        return builder.build();
-    }
-
-    protected static List<String> getSystemPropertyList(String property) {
-        String value = System.getProperty(property);
-        return value == null
-                ? Collections.emptyList()
-                : Arrays.asList(value.split(","));
-    }
-
-}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/ServerConfigProvider.java b/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/ServerConfigProvider.java
index c66305f..485f215 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/ServerConfigProvider.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/platform/serverconfig/ServerConfigProvider.java
@@ -20,37 +20,57 @@
 
 package me.lucko.spark.common.platform.serverconfig;
 
+import com.google.common.collect.ImmutableMap;
 import com.google.gson.JsonElement;
 
+import me.lucko.spark.common.platform.MetadataProvider;
+
+import java.util.Arrays;
+import java.util.Collection;
 import java.util.Collections;
-import java.util.LinkedHashMap;
+import java.util.List;
 import java.util.Map;
 
 /**
- * Function to export server configuration files for access within the spark viewer.
+ * Abstract implementation of {@link MetadataProvider} which
+ * provides server configuration data.
+ *
+ * <p>This implementation is able to delete hidden paths from
+ * the configurations before they are sent to the viewer.</p>
  */
-@FunctionalInterface
-public interface ServerConfigProvider {
-
-    /**
-     * Loads a map of the server configuration files.
-     *
-     * <p>The key is the name of the file and the value is a
-     * {@link JsonElement} of the contents.</p>
-     *
-     * @return the exported server configurations
-     */
-    Map<String, JsonElement> loadServerConfigurations();
-
-    default Map<String, String> exportServerConfigurations() {
-        Map<String, String> map = new LinkedHashMap<>();
-        loadServerConfigurations().forEach((key, value) -> map.put(key, value.toString()));
-        return map;
+public abstract class ServerConfigProvider implements MetadataProvider {
+    private final Map<String, ConfigParser> files;
+    private final ExcludedConfigFilter hiddenPathFilters;
+
+    protected ServerConfigProvider(Map<String, ConfigParser> files, Collection<String> hiddenPaths) {
+        this.files = files;
+        this.hiddenPathFilters = new ExcludedConfigFilter(hiddenPaths);
+    }
+
+    @Override
+    public final Map<String, JsonElement> get() {
+        ImmutableMap.Builder<String, JsonElement> builder = ImmutableMap.builder();
+
+        this.files.forEach((path, parser) -> {
+            try {
+                JsonElement json = parser.load(path, this.hiddenPathFilters);
+                if (json == null) {
+                    return;
+                }
+                builder.put(path, json);
+            } catch (Exception e) {
+                e.printStackTrace();
+            }
+        });
+
+        return builder.build();
     }
 
-    /**
-     * A no-op implementation
-     */
-    ServerConfigProvider NO_OP = Collections::emptyMap;
+    protected static List<String> getSystemPropertyList(String property) {
+        String value = System.getProperty(property);
+        return value == null
+                ? Collections.emptyList()
+                : Arrays.asList(value.split(","));
+    }
 
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/platform/world/AsyncWorldInfoProvider.java b/spark-common/src/main/java/me/lucko/spark/common/platform/world/AsyncWorldInfoProvider.java
new file mode 100644
index 0000000..82cddef
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/platform/world/AsyncWorldInfoProvider.java
@@ -0,0 +1,90 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.platform.world;
+
+import me.lucko.spark.common.SparkPlatform;
+import me.lucko.spark.common.SparkPlugin;
+
+import java.util.concurrent.CompletableFuture;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.TimeoutException;
+import java.util.function.Function;
+import java.util.logging.Level;
+
+/**
+ * Async-friendly wrapper around {@link WorldInfoProvider}.
+ */
+public class AsyncWorldInfoProvider {
+    private static final int TIMEOUT_SECONDS = 5;
+
+    private final SparkPlatform platform;
+    private final WorldInfoProvider provider;
+
+    public AsyncWorldInfoProvider(SparkPlatform platform, WorldInfoProvider provider) {
+        this.platform = platform;
+        this.provider = provider == WorldInfoProvider.NO_OP ? null : provider;
+    }
+
+    private <T> CompletableFuture<T> async(Function<WorldInfoProvider, T> function) {
+        if (this.provider == null) {
+            return null;
+        }
+
+        if (this.provider.mustCallSync()) {
+            SparkPlugin plugin = this.platform.getPlugin();
+            return CompletableFuture.supplyAsync(() -> function.apply(this.provider), plugin::executeSync);
+        } else {
+            return CompletableFuture.completedFuture(function.apply(this.provider));
+        }
+    }
+
+    private <T> T get(CompletableFuture<T> future) {
+        if (future == null) {
+            return null;
+        }
+
+        try {
+            return future.get(TIMEOUT_SECONDS, TimeUnit.SECONDS);
+        } catch (InterruptedException | ExecutionException e) {
+            throw new RuntimeException(e);
+        } catch (TimeoutException e) {
+            this.platform.getPlugin().log(Level.WARNING, "Timed out waiting for world statistics");
+            return null;
+        }
+    }
+
+    public CompletableFuture<WorldInfoProvider.CountsResult> pollCounts() {
+        return async(WorldInfoProvider::pollCounts);
+    }
+
+    public CompletableFuture<WorldInfoProvider.ChunksResult<? extends ChunkInfo<?>>> pollChunks() {
+        return async(WorldInfoProvider::pollChunks);
+    }
+
+    public WorldInfoProvider.CountsResult getCounts() {
+        return get(pollCounts());
+    }
+
+    public WorldInfoProvider.ChunksResult<? extends ChunkInfo<?>> getChunks() {
+        return get(pollChunks());
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldInfoProvider.java b/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldInfoProvider.java
index 9494816..7fb581d 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldInfoProvider.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldInfoProvider.java
@@ -29,20 +29,37 @@ import java.util.Map;
  */
 public interface WorldInfoProvider {
 
-    WorldInfoProvider NO_OP = () -> null;
+    WorldInfoProvider NO_OP = new WorldInfoProvider() {
+        @Override
+        public CountsResult pollCounts() {
+            return null;
+        }
+
+        @Override
+        public ChunksResult<? extends ChunkInfo<?>> pollChunks() {
+            return null;
+        }
+    };
+
+    /**
+     * Polls for counts.
+     *
+     * @return the counts
+     */
+    CountsResult pollCounts();
 
     /**
-     * Polls for information.
+     * Polls for chunk information.
      *
-     * @return the information
+     * @return the chunk information
      */
-    Result<? extends ChunkInfo<?>> poll();
+    ChunksResult<? extends ChunkInfo<?>> pollChunks();
 
     default boolean mustCallSync() {
         return true;
     }
 
-    final class Result<T> {
+    final class ChunksResult<T extends ChunkInfo<?>> {
         private final Map<String, List<T>> worlds = new HashMap<>();
 
         public void put(String worldName, List<T> chunks) {
@@ -54,4 +71,34 @@ public interface WorldInfoProvider {
         }
     }
 
+    final class CountsResult {
+        private final int players;
+        private final int entities;
+        private final int tileEntities;
+        private final int chunks;
+
+        public CountsResult(int players, int entities, int tileEntities, int chunks) {
+            this.players = players;
+            this.entities = entities;
+            this.tileEntities = tileEntities;
+            this.chunks = chunks;
+        }
+
+        public int players() {
+            return this.players;
+        }
+
+        public int entities() {
+            return this.entities;
+        }
+
+        public int tileEntities() {
+            return this.tileEntities;
+        }
+
+        public int chunks() {
+            return this.chunks;
+        }
+    }
+
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldStatisticsProvider.java b/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldStatisticsProvider.java
index 80c35a6..7e63222 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldStatisticsProvider.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/platform/world/WorldStatisticsProvider.java
@@ -20,8 +20,6 @@
 
 package me.lucko.spark.common.platform.world;
 
-import me.lucko.spark.common.SparkPlatform;
-import me.lucko.spark.common.SparkPlugin;
 import me.lucko.spark.proto.SparkProtos.WorldStatistics;
 
 import java.util.ArrayList;
@@ -30,46 +28,17 @@ import java.util.HashSet;
 import java.util.Iterator;
 import java.util.List;
 import java.util.Set;
-import java.util.concurrent.CompletableFuture;
-import java.util.concurrent.ExecutionException;
-import java.util.concurrent.TimeUnit;
-import java.util.concurrent.TimeoutException;
 import java.util.concurrent.atomic.AtomicInteger;
-import java.util.logging.Level;
 
 public class WorldStatisticsProvider {
-    private final SparkPlatform platform;
-    private final WorldInfoProvider provider;
+    private final AsyncWorldInfoProvider provider;
 
-    public WorldStatisticsProvider(SparkPlatform platform, WorldInfoProvider provider) {
-        this.platform = platform;
+    public WorldStatisticsProvider(AsyncWorldInfoProvider provider) {
         this.provider = provider;
     }
 
     public WorldStatistics getWorldStatistics() {
-        if (this.provider == WorldInfoProvider.NO_OP) {
-            return null;
-        }
-
-        CompletableFuture<WorldInfoProvider.Result<? extends ChunkInfo<?>>> future;
-
-        if (this.provider.mustCallSync()) {
-            SparkPlugin plugin = this.platform.getPlugin();
-            future = CompletableFuture.supplyAsync(this.provider::poll, plugin::executeSync);
-        } else {
-            future = CompletableFuture.completedFuture(this.provider.poll());
-        }
-
-        WorldInfoProvider.Result<? extends ChunkInfo<?>> result;
-        try {
-            result = future.get(5, TimeUnit.SECONDS);
-        } catch (InterruptedException | ExecutionException e) {
-            throw new RuntimeException(e);
-        } catch (TimeoutException e) {
-            this.platform.getPlugin().log(Level.WARNING, "Timed out waiting for world statistics");
-            return null;
-        }
-
+        WorldInfoProvider.ChunksResult<? extends ChunkInfo<?>> result = provider.getChunks();
         if (result == null) {
             return null;
         }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/AbstractSampler.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/AbstractSampler.java
index 1c217db..e324fd3 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/AbstractSampler.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/AbstractSampler.java
@@ -23,17 +23,22 @@ package me.lucko.spark.common.sampler;
 import me.lucko.spark.common.SparkPlatform;
 import me.lucko.spark.common.command.sender.CommandSender;
 import me.lucko.spark.common.monitor.memory.GarbageCollectorStatistics;
+import me.lucko.spark.common.platform.MetadataProvider;
 import me.lucko.spark.common.platform.serverconfig.ServerConfigProvider;
 import me.lucko.spark.common.sampler.aggregator.DataAggregator;
 import me.lucko.spark.common.sampler.node.MergeMode;
 import me.lucko.spark.common.sampler.node.ThreadNode;
-import me.lucko.spark.common.tick.TickHook;
-import me.lucko.spark.common.util.ClassSourceLookup;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
+import me.lucko.spark.common.sampler.source.SourceMetadata;
+import me.lucko.spark.common.sampler.window.ProtoTimeEncoder;
+import me.lucko.spark.common.sampler.window.WindowStatisticsCollector;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerData;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerMetadata;
 
+import java.util.Collection;
 import java.util.Comparator;
 import java.util.List;
+import java.util.Locale;
 import java.util.Map;
 import java.util.concurrent.CompletableFuture;
 
@@ -54,23 +59,28 @@ public abstract class AbstractSampler implements Sampler {
     /** The time when sampling first began */
     protected long startTime = -1;
 
-    /** The game tick when sampling first began */
-    protected int startTick = -1;
-
     /** The unix timestamp (in millis) when this sampler should automatically complete. */
     protected final long autoEndTime; // -1 for nothing
 
+    /** If the sampler is running in the background */
+    protected boolean background;
+
+    /** Collects statistics for each window in the sample */
+    protected final WindowStatisticsCollector windowStatisticsCollector;
+
     /** A future to encapsulate the completion of this sampler instance */
     protected final CompletableFuture<Sampler> future = new CompletableFuture<>();
 
     /** The garbage collector statistics when profiling started */
     protected Map<String, GarbageCollectorStatistics> initialGcStats;
 
-    protected AbstractSampler(SparkPlatform platform, int interval, ThreadDumper threadDumper, long autoEndTime) {
+    protected AbstractSampler(SparkPlatform platform, SamplerSettings settings) {
         this.platform = platform;
-        this.interval = interval;
-        this.threadDumper = threadDumper;
-        this.autoEndTime = autoEndTime;
+        this.interval = settings.interval();
+        this.threadDumper = settings.threadDumper();
+        this.autoEndTime = settings.autoEndTime();
+        this.background = settings.runningInBackground();
+        this.windowStatisticsCollector = new WindowStatisticsCollector(platform);
     }
 
     @Override
@@ -87,6 +97,11 @@ public abstract class AbstractSampler implements Sampler {
     }
 
     @Override
+    public boolean isRunningInBackground() {
+        return this.background;
+    }
+
+    @Override
     public CompletableFuture<Sampler> getFuture() {
         return this.future;
     }
@@ -102,11 +117,11 @@ public abstract class AbstractSampler implements Sampler {
     @Override
     public void start() {
         this.startTime = System.currentTimeMillis();
+    }
 
-        TickHook tickHook = this.platform.getTickHook();
-        if (tickHook != null) {
-            this.startTick = tickHook.getCurrentTick();
-        }
+    @Override
+    public void stop(boolean cancelled) {
+        this.windowStatisticsCollector.stop();
     }
 
     protected void writeMetadataToProto(SamplerData.Builder proto, SparkPlatform platform, CommandSender creator, String comment, DataAggregator dataAggregator) {
@@ -123,12 +138,9 @@ public abstract class AbstractSampler implements Sampler {
             metadata.setComment(comment);
         }
 
-        if (this.startTick != -1) {
-            TickHook tickHook = this.platform.getTickHook();
-            if (tickHook != null) {
-                int numberOfTicks = tickHook.getCurrentTick() - this.startTick;
-                metadata.setNumberOfTicks(numberOfTicks);
-            }
+        int totalTicks = this.windowStatisticsCollector.getTotalTicks();
+        if (totalTicks != -1) {
+            metadata.setNumberOfTicks(totalTicks);
         }
 
         try {
@@ -145,27 +157,60 @@ public abstract class AbstractSampler implements Sampler {
 
         try {
             ServerConfigProvider serverConfigProvider = platform.getPlugin().createServerConfigProvider();
-            metadata.putAllServerConfigurations(serverConfigProvider.exportServerConfigurations());
+            if (serverConfigProvider != null) {
+                metadata.putAllServerConfigurations(serverConfigProvider.export());
+            }
+        } catch (Exception e) {
+            e.printStackTrace();
+        }
+
+        try {
+            MetadataProvider extraMetadataProvider = platform.getPlugin().createExtraMetadataProvider();
+            if (extraMetadataProvider != null) {
+                metadata.putAllExtraPlatformMetadata(extraMetadataProvider.export());
+            }
         } catch (Exception e) {
             e.printStackTrace();
         }
 
+        Collection<SourceMetadata> knownSources = platform.getPlugin().getKnownSources();
+        for (SourceMetadata source : knownSources) {
+            metadata.putSources(source.getName().toLowerCase(Locale.ROOT), source.toProto());
+        }
+
         proto.setMetadata(metadata);
     }
 
-    protected void writeDataToProto(SamplerData.Builder proto, DataAggregator dataAggregator, Comparator<ThreadNode> outputOrder, MergeMode mergeMode, ClassSourceLookup classSourceLookup) {
+    protected void writeDataToProto(SamplerData.Builder proto, DataAggregator dataAggregator, MergeMode mergeMode, ClassSourceLookup classSourceLookup) {
         List<ThreadNode> data = dataAggregator.exportData();
-        data.sort(outputOrder);
+        data.sort(Comparator.comparing(ThreadNode::getThreadLabel));
 
         ClassSourceLookup.Visitor classSourceVisitor = ClassSourceLookup.createVisitor(classSourceLookup);
 
+        ProtoTimeEncoder timeEncoder = new ProtoTimeEncoder(data);
+        int[] timeWindows = timeEncoder.getKeys();
+        for (int timeWindow : timeWindows) {
+            proto.addTimeWindows(timeWindow);
+        }
+
+        this.windowStatisticsCollector.ensureHasStatisticsForAllWindows(timeWindows);
+        proto.putAllTimeWindowStatistics(this.windowStatisticsCollector.export());
+
         for (ThreadNode entry : data) {
-            proto.addThreads(entry.toProto(mergeMode));
+            proto.addThreads(entry.toProto(mergeMode, timeEncoder));
             classSourceVisitor.visit(entry);
         }
 
-        if (classSourceVisitor.hasMappings()) {
-            proto.putAllClassSources(classSourceVisitor.getMapping());
+        if (classSourceVisitor.hasClassSourceMappings()) {
+            proto.putAllClassSources(classSourceVisitor.getClassSourceMapping());
+        }
+
+        if (classSourceVisitor.hasMethodSourceMappings()) {
+            proto.putAllMethodSources(classSourceVisitor.getMethodSourceMapping());
+        }
+
+        if (classSourceVisitor.hasLineSourceMappings()) {
+            proto.putAllLineSources(classSourceVisitor.getLineSourceMapping());
         }
     }
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/BackgroundSamplerManager.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/BackgroundSamplerManager.java
new file mode 100644
index 0000000..7e3b6b4
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/BackgroundSamplerManager.java
@@ -0,0 +1,115 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler;
+
+import me.lucko.spark.common.SparkPlatform;
+import me.lucko.spark.common.platform.PlatformInfo;
+import me.lucko.spark.common.util.Configuration;
+
+import java.util.logging.Level;
+
+public class BackgroundSamplerManager {
+
+    private static final String OPTION_ENABLED = "backgroundProfiler";
+    private static final String OPTION_ENGINE = "backgroundProfilerEngine";
+    private static final String OPTION_INTERVAL = "backgroundProfilerInterval";
+
+    private static final String MARKER_FAILED = "_marker_background_profiler_failed";
+
+    private final SparkPlatform platform;
+    private final Configuration configuration;
+    private final boolean enabled;
+
+    public BackgroundSamplerManager(SparkPlatform platform, Configuration configuration) {
+        this.platform = platform;
+        this.configuration = configuration;
+
+        PlatformInfo.Type type = this.platform.getPlugin().getPlatformInfo().getType();
+        this.enabled = type != PlatformInfo.Type.CLIENT && this.configuration.getBoolean(OPTION_ENABLED, type == PlatformInfo.Type.SERVER);
+    }
+
+    public void initialise() {
+        if (!this.enabled) {
+            return;
+        }
+
+        // are we enabling the background profiler by default for the first time?
+        boolean didEnableByDefault = false;
+        if (!this.configuration.contains(OPTION_ENABLED)) {
+            this.configuration.setBoolean(OPTION_ENABLED, true);
+            didEnableByDefault = true;
+        }
+
+        // did the background profiler fail to start on the previous attempt?
+        if (this.configuration.getBoolean(MARKER_FAILED, false)) {
+            this.platform.getPlugin().log(Level.WARNING, "It seems the background profiler failed to start when spark was last enabled. Sorry about that!");
+            this.platform.getPlugin().log(Level.WARNING, "In the future, spark will try to use the built-in Java profiling engine instead.");
+
+            this.configuration.remove(MARKER_FAILED);
+            this.configuration.setString(OPTION_ENGINE, "java");
+            this.configuration.save();
+        }
+
+        this.platform.getPlugin().log(Level.INFO, "Starting background profiler...");
+
+        if (didEnableByDefault) {
+            // set the failed marker and save before we try to start the profiler,
+            // then remove the marker afterwards if everything goes ok!
+            this.configuration.setBoolean(MARKER_FAILED, true);
+            this.configuration.save();
+        }
+
+        try {
+            startSampler();
+
+            if (didEnableByDefault) {
+                this.configuration.remove(MARKER_FAILED);
+                this.configuration.save();
+            }
+
+        } catch (Throwable e) {
+            e.printStackTrace();
+        }
+    }
+
+    public boolean restartBackgroundSampler() {
+        if (this.enabled) {
+            startSampler();
+            return true;
+        }
+        return false;
+    }
+
+    private void startSampler() {
+        boolean forceJavaEngine = this.configuration.getString(OPTION_ENGINE, "async").equals("java");
+
+        Sampler sampler = new SamplerBuilder()
+                .background(true)
+                .threadDumper(this.platform.getPlugin().getDefaultThreadDumper())
+                .threadGrouper(ThreadGrouper.BY_POOL)
+                .samplingInterval(this.configuration.getInteger(OPTION_INTERVAL, 10))
+                .forceJavaSampler(forceJavaEngine)
+                .start(this.platform);
+
+        this.platform.getSamplerContainer().setActiveSampler(sampler);
+    }
+
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/Sampler.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/Sampler.java
index 84f2da1..36a63f1 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/Sampler.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/Sampler.java
@@ -23,11 +23,9 @@ package me.lucko.spark.common.sampler;
 import me.lucko.spark.common.SparkPlatform;
 import me.lucko.spark.common.command.sender.CommandSender;
 import me.lucko.spark.common.sampler.node.MergeMode;
-import me.lucko.spark.common.sampler.node.ThreadNode;
-import me.lucko.spark.common.util.ClassSourceLookup;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerData;
 
-import java.util.Comparator;
 import java.util.concurrent.CompletableFuture;
 
 /**
@@ -43,7 +41,7 @@ public interface Sampler {
     /**
      * Stops the sampler.
      */
-    void stop();
+    void stop(boolean cancelled);
 
     /**
      * Gets the time when the sampler started (unix timestamp in millis)
@@ -60,6 +58,13 @@ public interface Sampler {
     long getAutoEndTime();
 
     /**
+     * If this sampler is running in the background. (wasn't started by a specific user)
+     *
+     * @return true if the sampler is running in the background
+     */
+    boolean isRunningInBackground();
+
+    /**
      * Gets a future to encapsulate the completion of the sampler
      *
      * @return a future
@@ -67,6 +72,6 @@ public interface Sampler {
     CompletableFuture<Sampler> getFuture();
 
     // Methods used to export the sampler data to the web viewer.
-    SamplerData toProto(SparkPlatform platform, CommandSender creator, Comparator<ThreadNode> outputOrder, String comment, MergeMode mergeMode, ClassSourceLookup classSourceLookup);
+    SamplerData toProto(SparkPlatform platform, CommandSender creator, String comment, MergeMode mergeMode, ClassSourceLookup classSourceLookup);
 
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerBuilder.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerBuilder.java
index 88b9d91..ec635ef 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerBuilder.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerBuilder.java
@@ -38,7 +38,8 @@ public class SamplerBuilder {
     private boolean ignoreSleeping = false;
     private boolean ignoreNative = false;
     private boolean useAsyncProfiler = true;
-    private long timeout = -1;
+    private long autoEndTime = -1;
+    private boolean background = false;
     private ThreadDumper threadDumper = ThreadDumper.ALL;
     private ThreadGrouper threadGrouper = ThreadGrouper.BY_NAME;
 
@@ -57,7 +58,12 @@ public class SamplerBuilder {
         if (timeout <= 0) {
             throw new IllegalArgumentException("timeout > 0");
         }
-        this.timeout = System.currentTimeMillis() + unit.toMillis(timeout);
+        this.autoEndTime = System.currentTimeMillis() + unit.toMillis(timeout);
+        return this;
+    }
+
+    public SamplerBuilder background(boolean background) {
+        this.background = background;
         return this;
     }
 
@@ -93,15 +99,24 @@ public class SamplerBuilder {
     }
 
     public Sampler start(SparkPlatform platform) {
+        boolean onlyTicksOverMode = this.ticksOver != -1 && this.tickHook != null;
+        boolean canUseAsyncProfiler = this.useAsyncProfiler &&
+                !onlyTicksOverMode &&
+                !(this.ignoreSleeping || this.ignoreNative) &&
+                !(this.threadDumper instanceof ThreadDumper.Regex) &&
+                AsyncProfilerAccess.getInstance(platform).checkSupported(platform);
+
+
         int intervalMicros = (int) (this.samplingInterval * 1000d);
+        SamplerSettings settings = new SamplerSettings(intervalMicros, this.threadDumper, this.threadGrouper, this.autoEndTime, this.background);
 
         Sampler sampler;
-        if (this.ticksOver != -1 && this.tickHook != null) {
-            sampler = new JavaSampler(platform, intervalMicros, this.threadDumper, this.threadGrouper, this.timeout, this.ignoreSleeping, this.ignoreNative, this.tickHook, this.ticksOver);
-        } else if (this.useAsyncProfiler && !(this.threadDumper instanceof ThreadDumper.Regex) && AsyncProfilerAccess.INSTANCE.checkSupported(platform)) {
-            sampler = new AsyncSampler(platform, intervalMicros, this.threadDumper, this.threadGrouper, this.timeout);
+        if (canUseAsyncProfiler) {
+            sampler = new AsyncSampler(platform, settings);
+        } else if (onlyTicksOverMode) {
+            sampler = new JavaSampler(platform, settings, this.ignoreSleeping, this.ignoreNative, this.tickHook, this.ticksOver);
         } else {
-            sampler = new JavaSampler(platform, intervalMicros, this.threadDumper, this.threadGrouper, this.timeout, this.ignoreSleeping, this.ignoreNative);
+            sampler = new JavaSampler(platform, settings, this.ignoreSleeping, this.ignoreNative);
         }
 
         sampler.start();
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerContainer.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerContainer.java
new file mode 100644
index 0000000..15b1029
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerContainer.java
@@ -0,0 +1,76 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler;
+
+import java.util.concurrent.atomic.AtomicReference;
+
+/**
+ * Container for the active sampler.
+ */
+public class SamplerContainer implements AutoCloseable {
+
+    private final AtomicReference<Sampler> activeSampler = new AtomicReference<>();
+
+    /**
+     * Gets the active sampler, or null if a sampler is not active.
+     *
+     * @return the active sampler
+     */
+    public Sampler getActiveSampler() {
+        return this.activeSampler.get();
+    }
+
+    /**
+     * Sets the active sampler, throwing an exception if another sampler is already active.
+     *
+     * @param sampler the sampler
+     */
+    public void setActiveSampler(Sampler sampler) {
+        if (!this.activeSampler.compareAndSet(null, sampler)) {
+            throw new IllegalStateException("Attempted to set active sampler when another was already active!");
+        }
+    }
+
+    /**
+     * Unsets the active sampler, if the provided sampler is active.
+     *
+     * @param sampler the sampler
+     */
+    public void unsetActiveSampler(Sampler sampler) {
+        this.activeSampler.compareAndSet(sampler, null);
+    }
+
+    /**
+     * Stops the active sampler, if there is one.
+     */
+    public void stopActiveSampler(boolean cancelled) {
+        Sampler sampler = this.activeSampler.getAndSet(null);
+        if (sampler != null) {
+            sampler.stop(cancelled);
+        }
+    }
+
+    @Override
+    public void close() {
+        stopActiveSampler(true);
+    }
+
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerSettings.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerSettings.java
new file mode 100644
index 0000000..6e55a43
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/SamplerSettings.java
@@ -0,0 +1,61 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler;
+
+/**
+ * Base settings for all samplers
+ */
+public class SamplerSettings {
+
+    private final int interval;
+    private final ThreadDumper threadDumper;
+    private final ThreadGrouper threadGrouper;
+    private final long autoEndTime;
+    private final boolean runningInBackground;
+
+    public SamplerSettings(int interval, ThreadDumper threadDumper, ThreadGrouper threadGrouper, long autoEndTime, boolean runningInBackground) {
+        this.interval = interval;
+        this.threadDumper = threadDumper;
+        this.threadGrouper = threadGrouper;
+        this.autoEndTime = autoEndTime;
+        this.runningInBackground = runningInBackground;
+    }
+
+    public int interval() {
+        return this.interval;
+    }
+
+    public ThreadDumper threadDumper() {
+        return this.threadDumper;
+    }
+
+    public ThreadGrouper threadGrouper() {
+        return this.threadGrouper;
+    }
+
+    public long autoEndTime() {
+        return this.autoEndTime;
+    }
+
+    public boolean runningInBackground() {
+        return this.runningInBackground;
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/AbstractDataAggregator.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/AbstractDataAggregator.java
index ad9dee4..2c003e5 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/AbstractDataAggregator.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/AbstractDataAggregator.java
@@ -27,6 +27,7 @@ import java.util.ArrayList;
 import java.util.List;
 import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
+import java.util.function.IntPredicate;
 
 /**
  * Abstract implementation of {@link DataAggregator}.
@@ -52,6 +53,11 @@ public abstract class AbstractDataAggregator implements DataAggregator {
     }
 
     @Override
+    public void pruneData(IntPredicate timeWindowPredicate) {
+        this.threadData.values().removeIf(node -> node.removeTimeWindowsRecursively(timeWindowPredicate));
+    }
+
+    @Override
     public List<ThreadNode> exportData() {
         List<ThreadNode> data = new ArrayList<>(this.threadData.values());
         for (ThreadNode node : data) {
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/DataAggregator.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/DataAggregator.java
index 5590a96..ed33204 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/DataAggregator.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/aggregator/DataAggregator.java
@@ -24,6 +24,7 @@ import me.lucko.spark.common.sampler.node.ThreadNode;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerMetadata;
 
 import java.util.List;
+import java.util.function.IntPredicate;
 
 /**
  * Aggregates sampling data.
@@ -38,6 +39,13 @@ public interface DataAggregator {
     List<ThreadNode> exportData();
 
     /**
+     * Prunes windows of data from this aggregator if the given {@code timeWindowPredicate} returns true.
+     *
+     * @param timeWindowPredicate the predicate
+     */
+    void pruneData(IntPredicate timeWindowPredicate);
+
+    /**
      * Gets metadata about the data aggregator instance.
      */
     SamplerMetadata.DataAggregator getMetadata();
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncDataAggregator.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncDataAggregator.java
index 3de3943..402330a 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncDataAggregator.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncDataAggregator.java
@@ -47,10 +47,10 @@ public class AsyncDataAggregator extends AbstractDataAggregator {
                 .build();
     }
 
-    public void insertData(ProfileSegment element) {
+    public void insertData(ProfileSegment element, int window) {
         try {
             ThreadNode node = getNode(this.threadGrouper.getGroup(element.getNativeThreadId(), element.getThreadName()));
-            node.log(STACK_TRACE_DESCRIBER, element.getStackTrace(), element.getTime());
+            node.log(STACK_TRACE_DESCRIBER, element.getStackTrace(), element.getTime(), window);
         } catch (Exception e) {
             e.printStackTrace();
         }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerAccess.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerAccess.java
index ef2c035..1480650 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerAccess.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerAccess.java
@@ -22,9 +22,9 @@ package me.lucko.spark.common.sampler.async;
 
 import com.google.common.collect.ImmutableTable;
 import com.google.common.collect.Table;
+import com.google.common.io.ByteStreams;
 
 import me.lucko.spark.common.SparkPlatform;
-import me.lucko.spark.common.util.TemporaryFiles;
 
 import one.profiler.AsyncProfiler;
 import one.profiler.Events;
@@ -32,19 +32,29 @@ import one.profiler.Events;
 import java.io.BufferedReader;
 import java.io.InputStream;
 import java.io.InputStreamReader;
+import java.io.OutputStream;
 import java.net.URL;
 import java.nio.file.Files;
 import java.nio.file.Path;
-import java.nio.file.StandardCopyOption;
 import java.util.Locale;
+import java.util.Objects;
 import java.util.logging.Level;
 import java.util.stream.Collectors;
 
 /**
  * Provides a bridge between spark and async-profiler.
  */
-public enum AsyncProfilerAccess {
-    INSTANCE;
+public class AsyncProfilerAccess {
+    private static AsyncProfilerAccess instance;
+
+    // singleton, needs a SparkPlatform for first init
+    public static synchronized AsyncProfilerAccess getInstance(SparkPlatform platform) {
+        if (instance == null) {
+            Objects.requireNonNull(platform, "platform");
+            instance = new AsyncProfilerAccess(platform);
+        }
+        return instance;
+    }
 
     /** An instance of the async-profiler Java API. */
     private final AsyncProfiler profiler;
@@ -55,13 +65,13 @@ public enum AsyncProfilerAccess {
     /** If profiler is null, contains the reason why setup failed */
     private final Exception setupException;
 
-    AsyncProfilerAccess() {
+    AsyncProfilerAccess(SparkPlatform platform) {
         AsyncProfiler profiler;
         ProfilingEvent profilingEvent = null;
         Exception setupException = null;
 
         try {
-            profiler = load();
+            profiler = load(platform);
             if (isEventSupported(profiler, ProfilingEvent.CPU, false)) {
                 profilingEvent = ProfilingEvent.CPU;
             } else if (isEventSupported(profiler, ProfilingEvent.WALL, true)) {
@@ -77,11 +87,11 @@ public enum AsyncProfilerAccess {
         this.setupException = setupException;
     }
 
-    public AsyncProfiler getProfiler() {
+    public AsyncProfilerJob startNewProfilerJob() {
         if (this.profiler == null) {
             throw new UnsupportedOperationException("async-profiler not supported", this.setupException);
         }
-        return this.profiler;
+        return AsyncProfilerJob.createNew(this, this.profiler);
     }
 
     public ProfilingEvent getProfilingEvent() {
@@ -106,7 +116,7 @@ public enum AsyncProfilerAccess {
         return this.profiler != null;
     }
 
-    private static AsyncProfiler load() throws Exception {
+    private static AsyncProfiler load(SparkPlatform platform) throws Exception {
         // check compatibility
         String os = System.getProperty("os.name").toLowerCase(Locale.ROOT).replace(" ", "");
         String arch = System.getProperty("os.arch").toLowerCase(Locale.ROOT);
@@ -135,10 +145,10 @@ public enum AsyncProfilerAccess {
             throw new IllegalStateException("Could not find " + resource + " in spark jar file");
         }
 
-        Path extractPath = TemporaryFiles.create("spark-", "-libasyncProfiler.so.tmp");
+        Path extractPath = platform.getTemporaryFiles().create("spark-", "-libasyncProfiler.so.tmp");
 
-        try (InputStream in = profilerResource.openStream()) {
-            Files.copy(in, extractPath, StandardCopyOption.REPLACE_EXISTING);
+        try (InputStream in = profilerResource.openStream(); OutputStream out = Files.newOutputStream(extractPath)) {
+            ByteStreams.copy(in, out);
         }
 
         // get an instance of async-profiler
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerJob.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerJob.java
new file mode 100644
index 0000000..d74b75f
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncProfilerJob.java
@@ -0,0 +1,276 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.async;
+
+import me.lucko.spark.common.SparkPlatform;
+import me.lucko.spark.common.sampler.ThreadDumper;
+import me.lucko.spark.common.sampler.async.jfr.JfrReader;
+
+import one.profiler.AsyncProfiler;
+
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.util.List;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicReference;
+import java.util.function.Predicate;
+
+/**
+ * Represents a profiling job within async-profiler.
+ *
+ * <p>Only one job can be running at a time. This is guarded by
+ * {@link #createNew(AsyncProfilerAccess, AsyncProfiler)}.</p>
+ */
+public class AsyncProfilerJob {
+
+    /**
+     * The currently active job.
+     */
+    private static final AtomicReference<AsyncProfilerJob> ACTIVE = new AtomicReference<>();
+
+    /**
+     * Creates a new {@link AsyncProfilerJob}.
+     *
+     * <p>Will throw an {@link IllegalStateException} if another job is already active.</p>
+     *
+     * @param access the profiler access object
+     * @param profiler the profiler
+     * @return the job
+     */
+    static AsyncProfilerJob createNew(AsyncProfilerAccess access, AsyncProfiler profiler) {
+        synchronized (ACTIVE) {
+            AsyncProfilerJob existing = ACTIVE.get();
+            if (existing != null) {
+                throw new IllegalStateException("Another profiler is already active: " + existing);
+            }
+
+            AsyncProfilerJob job = new AsyncProfilerJob(access, profiler);
+            ACTIVE.set(job);
+            return job;
+        }
+    }
+
+    /** The async-profiler access object */
+    private final AsyncProfilerAccess access;
+    /** The async-profiler instance */
+    private final AsyncProfiler profiler;
+
+    // Set on init
+    /** The platform */
+    private SparkPlatform platform;
+    /** The sampling interval in microseconds */
+    private int interval;
+    /** The thread dumper */
+    private ThreadDumper threadDumper;
+    /** The profiling window */
+    private int window;
+    /** If the profiler should run in quiet mode */
+    private boolean quiet;
+
+    /** The file used by async-profiler to output data */
+    private Path outputFile;
+
+    private AsyncProfilerJob(AsyncProfilerAccess access, AsyncProfiler profiler) {
+        this.access = access;
+        this.profiler = profiler;
+    }
+
+    /**
+     * Executes an async-profiler command.
+     *
+     * @param command the command
+     * @return the output
+     */
+    private String execute(String command) {
+        try {
+            return this.profiler.execute(command);
+        } catch (IOException e) {
+            throw new RuntimeException("Exception whilst executing profiler command", e);
+        }
+    }
+
+    /**
+     * Checks to ensure that this job is still active.
+     */
+    private void checkActive() {
+        if (ACTIVE.get() != this) {
+            throw new IllegalStateException("Profiler job no longer active!");
+        }
+    }
+
+    // Initialise the job
+    public void init(SparkPlatform platform, int interval, ThreadDumper threadDumper, int window, boolean quiet) {
+        this.platform = platform;
+        this.interval = interval;
+        this.threadDumper = threadDumper;
+        this.window = window;
+        this.quiet = quiet;
+    }
+
+    /**
+     * Starts the job.
+     */
+    public void start() {
+        checkActive();
+
+        try {
+            // create a new temporary output file
+            try {
+                this.outputFile = this.platform.getTemporaryFiles().create("spark-", "-profile-data.jfr.tmp");
+            } catch (IOException e) {
+                throw new RuntimeException("Unable to create temporary output file", e);
+            }
+
+            // construct a command to send to async-profiler
+            String command = "start,event=" + this.access.getProfilingEvent() + ",interval=" + this.interval + "us,threads,jfr,file=" + this.outputFile.toString();
+            if (this.quiet) {
+                command += ",loglevel=NONE";
+            }
+            if (this.threadDumper instanceof ThreadDumper.Specific) {
+                command += ",filter";
+            }
+
+            // start the profiler
+            String resp = execute(command).trim();
+
+            if (!resp.equalsIgnoreCase("profiling started")) {
+                throw new RuntimeException("Unexpected response: " + resp);
+            }
+
+            // append threads to be profiled, if necessary
+            if (this.threadDumper instanceof ThreadDumper.Specific) {
+                ThreadDumper.Specific threadDumper = (ThreadDumper.Specific) this.threadDumper;
+                for (Thread thread : threadDumper.getThreads()) {
+                    this.profiler.addThread(thread);
+                }
+            }
+
+        } catch (Exception e) {
+            try {
+                this.profiler.stop();
+            } catch (Exception e2) {
+                // ignore
+            }
+            close();
+
+            throw e;
+        }
+    }
+
+    /**
+     * Stops the job.
+     */
+    public void stop() {
+        checkActive();
+
+        try {
+            this.profiler.stop();
+        } catch (IllegalStateException e) {
+            if (!e.getMessage().equals("Profiler is not active")) { // ignore
+                throw e;
+            }
+        } finally {
+            close();
+        }
+    }
+
+    /**
+     * Aggregates the collected data.
+     */
+    public void aggregate(AsyncDataAggregator dataAggregator) {
+
+        Predicate<String> threadFilter;
+        if (this.threadDumper instanceof ThreadDumper.Specific) {
+            ThreadDumper.Specific specificDumper = (ThreadDumper.Specific) this.threadDumper;
+            threadFilter = n -> specificDumper.getThreadNames().contains(n.toLowerCase());
+        } else {
+            threadFilter = n -> true;
+        }
+
+        // read the jfr file produced by async-profiler
+        try (JfrReader reader = new JfrReader(this.outputFile)) {
+            readSegments(reader, threadFilter, dataAggregator, this.window);
+        } catch (Exception e) {
+            boolean fileExists;
+            try {
+                fileExists = Files.exists(this.outputFile) && Files.size(this.outputFile) != 0;
+            } catch (IOException ex) {
+                fileExists = false;
+            }
+
+            if (fileExists) {
+                throw new JfrParsingException("Error parsing JFR data from profiler output", e);
+            } else {
+                throw new JfrParsingException("Error parsing JFR data from profiler output - file " + this.outputFile + " does not exist!", e);
+            }
+        }
+
+        deleteOutputFile();
+    }
+
+    public void deleteOutputFile() {
+        try {
+            Files.deleteIfExists(this.outputFile);
+        } catch (IOException e) {
+            // ignore
+        }
+    }
+
+    private void readSegments(JfrReader reader, Predicate<String> threadFilter, AsyncDataAggregator dataAggregator, int window) throws IOException {
+        List<JfrReader.ExecutionSample> samples = reader.readAllEvents(JfrReader.ExecutionSample.class);
+        for (int i = 0; i < samples.size(); i++) {
+            JfrReader.ExecutionSample sample = samples.get(i);
+
+            long duration;
+            if (i == 0) {
+                // we don't really know the duration of the first sample, so just use the sampling
+                // interval
+                duration = this.interval;
+            } else {
+                // calculate the duration of the sample by calculating the time elapsed since the
+                // previous sample
+                duration = TimeUnit.NANOSECONDS.toMicros(sample.time - samples.get(i - 1).time);
+            }
+
+            String threadName = reader.threads.get((long) sample.tid);
+            if (threadName == null) {
+                continue;
+            }
+
+            if (!threadFilter.test(threadName)) {
+                continue;
+            }
+
+            // parse the segment and give it to the data aggregator
+            ProfileSegment segment = ProfileSegment.parseSegment(reader, sample, threadName, duration);
+            dataAggregator.insertData(segment, window);
+        }
+    }
+
+    public int getWindow() {
+        return this.window;
+    }
+
+    private void close() {
+        ACTIVE.compareAndSet(this, null);
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncSampler.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncSampler.java
index d8288da..178f055 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncSampler.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/AsyncSampler.java
@@ -25,64 +25,43 @@ import com.google.common.util.concurrent.ThreadFactoryBuilder;
 import me.lucko.spark.common.SparkPlatform;
 import me.lucko.spark.common.command.sender.CommandSender;
 import me.lucko.spark.common.sampler.AbstractSampler;
-import me.lucko.spark.common.sampler.ThreadDumper;
-import me.lucko.spark.common.sampler.ThreadGrouper;
-import me.lucko.spark.common.sampler.async.jfr.JfrReader;
+import me.lucko.spark.common.sampler.SamplerSettings;
 import me.lucko.spark.common.sampler.node.MergeMode;
-import me.lucko.spark.common.sampler.node.ThreadNode;
-import me.lucko.spark.common.util.ClassSourceLookup;
-import me.lucko.spark.common.util.TemporaryFiles;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
+import me.lucko.spark.common.sampler.window.ProfilingWindowUtils;
+import me.lucko.spark.common.tick.TickHook;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerData;
 
-import one.profiler.AsyncProfiler;
-
-import java.io.IOException;
-import java.nio.charset.StandardCharsets;
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.Comparator;
-import java.util.List;
 import java.util.concurrent.Executors;
 import java.util.concurrent.ScheduledExecutorService;
 import java.util.concurrent.TimeUnit;
-import java.util.function.Predicate;
+import java.util.function.IntPredicate;
 
 /**
  * A sampler implementation using async-profiler.
  */
 public class AsyncSampler extends AbstractSampler {
-    private final AsyncProfiler profiler;
+    private final AsyncProfilerAccess profilerAccess;
 
     /** Responsible for aggregating and then outputting collected sampling data */
     private final AsyncDataAggregator dataAggregator;
 
-    /** Flag to mark if the output has been completed */
-    private boolean outputComplete = false;
-
-    /** The temporary output file */
-    private Path outputFile;
+    /** Mutex for the current profiler job */
+    private final Object[] currentJobMutex = new Object[0];
 
-    /** The executor used for timeouts */
-    private ScheduledExecutorService timeoutExecutor;
+    /** Current profiler job */
+    private AsyncProfilerJob currentJob;
 
-    public AsyncSampler(SparkPlatform platform, int interval, ThreadDumper threadDumper, ThreadGrouper threadGrouper, long endTime) {
-        super(platform, interval, threadDumper, endTime);
-        this.profiler = AsyncProfilerAccess.INSTANCE.getProfiler();
-        this.dataAggregator = new AsyncDataAggregator(threadGrouper);
-    }
+    /** The executor used for scheduling and management */
+    private ScheduledExecutorService scheduler;
 
-    /**
-     * Executes a profiler command.
-     *
-     * @param command the command to execute
-     * @return the response
-     */
-    private String execute(String command) {
-        try {
-            return this.profiler.execute(command);
-        } catch (IOException e) {
-            throw new RuntimeException("Exception whilst executing profiler command", e);
-        }
+    public AsyncSampler(SparkPlatform platform, SamplerSettings settings) {
+        super(platform, settings);
+        this.profilerAccess = AsyncProfilerAccess.getInstance(platform);
+        this.dataAggregator = new AsyncDataAggregator(settings.threadGrouper());
+        this.scheduler = Executors.newSingleThreadScheduledExecutor(
+                new ThreadFactoryBuilder().setNameFormat("spark-asyncsampler-worker-thread").build()
+        );
     }
 
     /**
@@ -92,33 +71,68 @@ public class AsyncSampler extends AbstractSampler {
     public void start() {
         super.start();
 
-        try {
-            this.outputFile = TemporaryFiles.create("spark-profile-", ".jfr.tmp");
-        } catch (IOException e) {
-            throw new RuntimeException("Unable to create temporary output file", e);
+        TickHook tickHook = this.platform.getTickHook();
+        if (tickHook != null) {
+            this.windowStatisticsCollector.startCountingTicks(tickHook);
         }
 
-        String command = "start,event=" + AsyncProfilerAccess.INSTANCE.getProfilingEvent() + ",interval=" + this.interval + "us,threads,jfr,file=" + this.outputFile.toString();
-        if (this.threadDumper instanceof ThreadDumper.Specific) {
-            command += ",filter";
-        }
+        int window = ProfilingWindowUtils.windowNow();
 
-        String resp = execute(command).trim();
-        if (!resp.equalsIgnoreCase("profiling started")) {
-            throw new RuntimeException("Unexpected response: " + resp);
-        }
+        AsyncProfilerJob job = this.profilerAccess.startNewProfilerJob();
+        job.init(this.platform, this.interval, this.threadDumper, window, this.background);
+        job.start();
+        this.currentJob = job;
 
-        if (this.threadDumper instanceof ThreadDumper.Specific) {
-            ThreadDumper.Specific threadDumper = (ThreadDumper.Specific) this.threadDumper;
-            for (Thread thread : threadDumper.getThreads()) {
-                this.profiler.addThread(thread);
-            }
-        }
+        // rotate the sampler job to put data into a new window
+        this.scheduler.scheduleAtFixedRate(
+                this::rotateProfilerJob,
+                ProfilingWindowUtils.WINDOW_SIZE_SECONDS,
+                ProfilingWindowUtils.WINDOW_SIZE_SECONDS,
+                TimeUnit.SECONDS
+        );
 
         recordInitialGcStats();
         scheduleTimeout();
     }
 
+    private void rotateProfilerJob() {
+        try {
+            synchronized (this.currentJobMutex) {
+                AsyncProfilerJob previousJob = this.currentJob;
+                if (previousJob == null) {
+                    return;
+                }
+
+                try {
+                    // stop the previous job
+                    previousJob.stop();
+
+                    // collect statistics for the window
+                    this.windowStatisticsCollector.measureNow(previousJob.getWindow());
+                } catch (Exception e) {
+                    e.printStackTrace();
+                }
+
+                // start a new job
+                int window = previousJob.getWindow() + 1;
+                AsyncProfilerJob newJob = this.profilerAccess.startNewProfilerJob();
+                newJob.init(this.platform, this.interval, this.threadDumper, window, this.background);
+                newJob.start();
+                this.currentJob = newJob;
+
+                // aggregate the output of the previous job
+                previousJob.aggregate(this.dataAggregator);
+
+                // prune data older than the history size
+                IntPredicate predicate = ProfilingWindowUtils.keepHistoryBefore(window);
+                this.dataAggregator.pruneData(predicate);
+                this.windowStatisticsCollector.pruneStatistics(predicate);
+            }
+        } catch (Throwable e) {
+            e.printStackTrace();
+        }
+    }
+
     private void scheduleTimeout() {
         if (this.autoEndTime == -1) {
             return;
@@ -129,12 +143,8 @@ public class AsyncSampler extends AbstractSampler {
             return;
         }
 
-        this.timeoutExecutor = Executors.newSingleThreadScheduledExecutor(
-                new ThreadFactoryBuilder().setNameFormat("spark-asyncsampler-timeout-thread").build()
-        );
-
-        this.timeoutExecutor.schedule(() -> {
-            stop();
+        this.scheduler.schedule(() -> {
+            stop(false);
             this.future.complete(this);
         }, delay, TimeUnit.MILLISECONDS);
     }
@@ -143,129 +153,32 @@ public class AsyncSampler extends AbstractSampler {
      * Stops the profiler.
      */
     @Override
-    public void stop() {
-        try {
-            this.profiler.stop();
-        } catch (IllegalStateException e) {
-            if (!e.getMessage().equals("Profiler is not active")) { // ignore
-                throw e;
+    public void stop(boolean cancelled) {
+        super.stop(cancelled);
+
+        synchronized (this.currentJobMutex) {
+            this.currentJob.stop();
+            if (!cancelled) {
+                this.windowStatisticsCollector.measureNow(this.currentJob.getWindow());
+                this.currentJob.aggregate(this.dataAggregator);
+            } else {
+                this.currentJob.deleteOutputFile();
             }
+            this.currentJob = null;
         }
 
-
-        if (this.timeoutExecutor != null) {
-            this.timeoutExecutor.shutdown();
-            this.timeoutExecutor = null;
+        if (this.scheduler != null) {
+            this.scheduler.shutdown();
+            this.scheduler = null;
         }
     }
 
     @Override
-    public SamplerData toProto(SparkPlatform platform, CommandSender creator, Comparator<ThreadNode> outputOrder, String comment, MergeMode mergeMode, ClassSourceLookup classSourceLookup) {
+    public SamplerData toProto(SparkPlatform platform, CommandSender creator, String comment, MergeMode mergeMode, ClassSourceLookup classSourceLookup) {
         SamplerData.Builder proto = SamplerData.newBuilder();
         writeMetadataToProto(proto, platform, creator, comment, this.dataAggregator);
-        aggregateOutput();
-        writeDataToProto(proto, this.dataAggregator, outputOrder, mergeMode, classSourceLookup);
+        writeDataToProto(proto, this.dataAggregator, mergeMode, classSourceLookup);
         return proto.build();
     }
 
-    private void aggregateOutput() {
-        if (this.outputComplete) {
-            return;
-        }
-        this.outputComplete = true;
-
-        Predicate<String> threadFilter;
-        if (this.threadDumper instanceof ThreadDumper.Specific) {
-            ThreadDumper.Specific threadDumper = (ThreadDumper.Specific) this.threadDumper;
-            threadFilter = n -> threadDumper.getThreadNames().contains(n.toLowerCase());
-        } else {
-            threadFilter = n -> true;
-        }
-
-        // read the jfr file produced by async-profiler
-        try (JfrReader reader = new JfrReader(this.outputFile)) {
-            readSegments(reader, threadFilter);
-        } catch (IOException e) {
-            throw new RuntimeException("Read error", e);
-        }
-
-        // delete the output file after reading
-        try {
-            Files.deleteIfExists(this.outputFile);
-        } catch (IOException e) {
-            // ignore
-        }
-    }
-
-    private void readSegments(JfrReader reader, Predicate<String> threadFilter) throws IOException {
-        List<JfrReader.ExecutionSample> samples = reader.readAllEvents(JfrReader.ExecutionSample.class);
-        for (int i = 0; i < samples.size(); i++) {
-            JfrReader.ExecutionSample sample = samples.get(i);
-
-            long duration;
-            if (i == 0) {
-                // we don't really know the duration of the first sample, so just use the sampling
-                // interval
-                duration = this.interval;
-            } else {
-                // calculate the duration of the sample by calculating the time elapsed since the
-                // previous sample
-                duration = TimeUnit.NANOSECONDS.toMicros(sample.time - samples.get(i - 1).time);
-            }
-
-            String threadName = reader.threads.get(sample.tid);
-            if (!threadFilter.test(threadName)) {
-                continue;
-            }
-
-            // parse the segment and give it to the data aggregator
-            ProfileSegment segment = parseSegment(reader, sample, threadName, duration);
-            this.dataAggregator.insertData(segment);
-        }
-    }
-
-    private static ProfileSegment parseSegment(JfrReader reader, JfrReader.ExecutionSample sample, String threadName, long duration) {
-        JfrReader.StackTrace stackTrace = reader.stackTraces.get(sample.stackTraceId);
-        int len = stackTrace.methods.length;
-
-        AsyncStackTraceElement[] stack = new AsyncStackTraceElement[len];
-        for (int i = 0; i < len; i++) {
-            stack[i] = parseStackFrame(reader, stackTrace.methods[i]);
-        }
-
-        return new ProfileSegment(sample.tid, threadName, stack, duration);
-    }
-
-    private static AsyncStackTraceElement parseStackFrame(JfrReader reader, long methodId) {
-        AsyncStackTraceElement result = reader.stackFrames.get(methodId);
-        if (result != null) {
-            return result;
-        }
-
-        JfrReader.MethodRef methodRef = reader.methods.get(methodId);
-        JfrReader.ClassRef classRef = reader.classes.get(methodRef.cls);
-
-        byte[] className = reader.symbols.get(classRef.name);
-        byte[] methodName = reader.symbols.get(methodRef.name);
-
-        if (className == null || className.length == 0) {
-            // native call
-            result = new AsyncStackTraceElement(
-                    AsyncStackTraceElement.NATIVE_CALL,
-                    new String(methodName, StandardCharsets.UTF_8),
-                    null
-            );
-        } else {
-            // java method
-            byte[] methodDesc = reader.symbols.get(methodRef.sig);
-            result = new AsyncStackTraceElement(
-                    new String(className, StandardCharsets.UTF_8).replace('/', '.'),
-                    new String(methodName, StandardCharsets.UTF_8),
-                    new String(methodDesc, StandardCharsets.UTF_8)
-            );
-        }
-
-        reader.stackFrames.put(methodId, result);
-        return result;
-    }
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/JfrParsingException.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/JfrParsingException.java
new file mode 100644
index 0000000..6dab359
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/JfrParsingException.java
@@ -0,0 +1,27 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.async;
+
+public class JfrParsingException extends RuntimeException {
+    public JfrParsingException(String message, Throwable cause) {
+        super(message, cause);
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/ProfileSegment.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/ProfileSegment.java
index 154e6fe..26debaf 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/ProfileSegment.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/ProfileSegment.java
@@ -20,6 +20,10 @@
 
 package me.lucko.spark.common.sampler.async;
 
+import me.lucko.spark.common.sampler.async.jfr.JfrReader;
+
+import java.nio.charset.StandardCharsets;
+
 /**
  * Represents a profile "segment".
  *
@@ -58,4 +62,50 @@ public class ProfileSegment {
     public long getTime() {
         return this.time;
     }
+
+    public static ProfileSegment parseSegment(JfrReader reader, JfrReader.ExecutionSample sample, String threadName, long duration) {
+        JfrReader.StackTrace stackTrace = reader.stackTraces.get(sample.stackTraceId);
+        int len = stackTrace.methods.length;
+
+        AsyncStackTraceElement[] stack = new AsyncStackTraceElement[len];
+        for (int i = 0; i < len; i++) {
+            stack[i] = parseStackFrame(reader, stackTrace.methods[i]);
+        }
+
+        return new ProfileSegment(sample.tid, threadName, stack, duration);
+    }
+
+    private static AsyncStackTraceElement parseStackFrame(JfrReader reader, long methodId) {
+        AsyncStackTraceElement result = reader.stackFrames.get(methodId);
+        if (result != null) {
+            return result;
+        }
+
+        JfrReader.MethodRef methodRef = reader.methods.get(methodId);
+        JfrReader.ClassRef classRef = reader.classes.get(methodRef.cls);
+
+        byte[] className = reader.symbols.get(classRef.name);
+        byte[] methodName = reader.symbols.get(methodRef.name);
+
+        if (className == null || className.length == 0) {
+            // native call
+            result = new AsyncStackTraceElement(
+                    AsyncStackTraceElement.NATIVE_CALL,
+                    new String(methodName, StandardCharsets.UTF_8),
+                    null
+            );
+        } else {
+            // java method
+            byte[] methodDesc = reader.symbols.get(methodRef.sig);
+            result = new AsyncStackTraceElement(
+                    new String(className, StandardCharsets.UTF_8).replace('/', '.'),
+                    new String(methodName, StandardCharsets.UTF_8),
+                    new String(methodDesc, StandardCharsets.UTF_8)
+            );
+        }
+
+        reader.stackFrames.put(methodId, result);
+        return result;
+    }
+
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/Dictionary.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/Dictionary.java
index 23223a2..60f6543 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/Dictionary.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/Dictionary.java
@@ -37,6 +37,10 @@ public class Dictionary<T> {
         size = 0;
     }
 
+    public int size() {
+        return this.size;
+    }
+
     public void put(long key, T value) {
         if (key == 0) {
             throw new IllegalArgumentException("Zero key not allowed");
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/JfrReader.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/JfrReader.java
index e0cc4e9..ea4985e 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/JfrReader.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/async/jfr/JfrReader.java
@@ -51,7 +51,7 @@ public class JfrReader implements Closeable {
 
     public final Dictionary<JfrClass> types = new Dictionary<>();
     public final Map<String, JfrClass> typesByName = new HashMap<>();
-    public final Dictionary<String> threads = new Dictionary<>();
+    public final Map<Long, String> threads = new HashMap<>(); // spark
     public final Dictionary<ClassRef> classes = new Dictionary<>();
     public final Dictionary<byte[]> symbols = new Dictionary<>();
     public final Dictionary<MethodRef> methods = new Dictionary<>();
@@ -324,7 +324,7 @@ public class JfrReader implements Closeable {
     }
 
     private void readThreads(boolean hasGroup) {
-        int count = threads.preallocate(getVarint());
+        int count = getVarint(); //threads.preallocate(getVarint());
         for (int i = 0; i < count; i++) {
             long id = getVarlong();
             String osName = getString();
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaDataAggregator.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaDataAggregator.java
index cc530d6..c51ec05 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaDataAggregator.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaDataAggregator.java
@@ -66,10 +66,11 @@ public abstract class JavaDataAggregator extends AbstractDataAggregator {
      * Inserts sampling data into this aggregator
      *
      * @param threadInfo the thread info
+     * @param window the window
      */
-    public abstract void insertData(ThreadInfo threadInfo);
+    public abstract void insertData(ThreadInfo threadInfo, int window);
 
-    protected void writeData(ThreadInfo threadInfo) {
+    protected void writeData(ThreadInfo threadInfo, int window) {
         if (this.ignoreSleeping && isSleeping(threadInfo)) {
             return;
         }
@@ -79,7 +80,7 @@ public abstract class JavaDataAggregator extends AbstractDataAggregator {
 
         try {
             ThreadNode node = getNode(this.threadGrouper.getGroup(threadInfo.getThreadId(), threadInfo.getThreadName()));
-            node.log(STACK_TRACE_DESCRIBER, threadInfo.getStackTrace(), this.interval);
+            node.log(STACK_TRACE_DESCRIBER, threadInfo.getStackTrace(), this.interval, window);
         } catch (Exception e) {
             e.printStackTrace();
         }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaSampler.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaSampler.java
index 913faee..72a37e8 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaSampler.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/JavaSampler.java
@@ -25,23 +25,23 @@ import com.google.common.util.concurrent.ThreadFactoryBuilder;
 import me.lucko.spark.common.SparkPlatform;
 import me.lucko.spark.common.command.sender.CommandSender;
 import me.lucko.spark.common.sampler.AbstractSampler;
-import me.lucko.spark.common.sampler.ThreadDumper;
-import me.lucko.spark.common.sampler.ThreadGrouper;
+import me.lucko.spark.common.sampler.SamplerSettings;
 import me.lucko.spark.common.sampler.node.MergeMode;
-import me.lucko.spark.common.sampler.node.ThreadNode;
+import me.lucko.spark.common.sampler.source.ClassSourceLookup;
+import me.lucko.spark.common.sampler.window.ProfilingWindowUtils;
+import me.lucko.spark.common.sampler.window.WindowStatisticsCollector;
 import me.lucko.spark.common.tick.TickHook;
-import me.lucko.spark.common.util.ClassSourceLookup;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerData;
 
 import java.lang.management.ManagementFactory;
 import java.lang.management.ThreadInfo;
 import java.lang.management.ThreadMXBean;
-import java.util.Comparator;
 import java.util.concurrent.Executors;
 import java.util.concurrent.ScheduledExecutorService;
 import java.util.concurrent.ScheduledFuture;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicInteger;
+import java.util.function.IntPredicate;
 
 /**
  * A sampler implementation using Java (WarmRoast).
@@ -62,26 +62,47 @@ public class JavaSampler extends AbstractSampler implements Runnable {
 
     /** Responsible for aggregating and then outputting collected sampling data */
     private final JavaDataAggregator dataAggregator;
+
+    /** The last window that was profiled */
+    private final AtomicInteger lastWindow = new AtomicInteger();
     
-    public JavaSampler(SparkPlatform platform, int interval, ThreadDumper threadDumper, ThreadGrouper threadGrouper, long endTime, boolean ignoreSleeping, boolean ignoreNative) {
-        super(platform, interval, threadDumper, endTime);
-        this.dataAggregator = new SimpleDataAggregator(this.workerPool, threadGrouper, interval, ignoreSleeping, ignoreNative);
+    public JavaSampler(SparkPlatform platform, SamplerSettings settings, boolean ignoreSleeping, boolean ignoreNative) {
+        super(platform, settings);
+        this.dataAggregator = new SimpleDataAggregator(this.workerPool, settings.threadGrouper(), settings.interval(), ignoreSleeping, ignoreNative);
     }
 
-    public JavaSampler(SparkPlatform platform, int interval, ThreadDumper threadDumper, ThreadGrouper threadGrouper, long endTime, boolean ignoreSleeping, boolean ignoreNative, TickHook tickHook, int tickLengthThreshold) {
-        super(platform, interval, threadDumper, endTime);
-        this.dataAggregator = new TickedDataAggregator(this.workerPool, threadGrouper, interval, ignoreSleeping, ignoreNative, tickHook, tickLengthThreshold);
+    public JavaSampler(SparkPlatform platform, SamplerSettings settings, boolean ignoreSleeping, boolean ignoreNative, TickHook tickHook, int tickLengthThreshold) {
+        super(platform, settings);
+        this.dataAggregator = new TickedDataAggregator(this.workerPool, settings.threadGrouper(), settings.interval(), ignoreSleeping, ignoreNative, tickHook, tickLengthThreshold);
     }
 
     @Override
     public void start() {
         super.start();
+
+        TickHook tickHook = this.platform.getTickHook();
+        if (tickHook != null) {
+            if (this.dataAggregator instanceof TickedDataAggregator) {
+                WindowStatisticsCollector.ExplicitTickCounter counter = this.windowStatisticsCollector.startCountingTicksExplicit(tickHook);
+                ((TickedDataAggregator) this.dataAggregator).setTickCounter(counter);
+            } else {
+                this.windowStatisticsCollector.startCountingTicks(tickHook);
+            }
+        }
+
         this.task = this.workerPool.scheduleAtFixedRate(this, 0, this.interval, TimeUnit.MICROSECONDS);
     }
 
     @Override
-    public void stop() {
+    public void stop(boolean cancelled) {
+        super.stop(cancelled);
+
         this.task.cancel(false);
+
+        if (!cancelled) {
+            // collect statistics for the final window
+            this.windowStatisticsCollector.measureNow(this.lastWindow.get());
+        }
     }
 
     @Override
@@ -89,27 +110,30 @@ public class JavaSampler extends AbstractSampler implements Runnable {
         // this is effectively synchronized, the worker pool will not allow this task
         // to concurrently execute.
         try {
-            if (this.autoEndTime != -1 && this.autoEndTime <= System.currentTimeMillis()) {
+            long time = System.currentTimeMillis();
+
+            if (this.autoEndTime != -1 && this.autoEndTime <= time) {
+                stop(false);
                 this.future.complete(this);
-                stop();
                 return;
             }
 
+            int window = ProfilingWindowUtils.unixMillisToWindow(time);
             ThreadInfo[] threadDumps = this.threadDumper.dumpThreads(this.threadBean);
-            this.workerPool.execute(new InsertDataTask(this.dataAggregator, threadDumps));
+            this.workerPool.execute(new InsertDataTask(threadDumps, window));
         } catch (Throwable t) {
+            stop(false);
             this.future.completeExceptionally(t);
-            stop();
         }
     }
 
-    private static final class InsertDataTask implements Runnable {
-        private final JavaDataAggregator dataAggregator;
+    private final class InsertDataTask implements Runnable {
         private final ThreadInfo[] threadDumps;
+        private final int window;
 
-        InsertDataTask(JavaDataAggregator dataAggregator, ThreadInfo[] threadDumps) {
-            this.dataAggregator = dataAggregator;
+        InsertDataTask(ThreadInfo[] threadDumps, int window) {
             this.threadDumps = threadDumps;
+            this.window = window;
         }
 
         @Override
@@ -118,16 +142,29 @@ public class JavaSampler extends AbstractSampler implements Runnable {
                 if (threadInfo.getThreadName() == null || threadInfo.getStackTrace() == null) {
                     continue;
                 }
-                this.dataAggregator.insertData(threadInfo);
+                JavaSampler.this.dataAggregator.insertData(threadInfo, this.window);
+            }
+
+            // if we have just stepped over into a new window...
+            int previousWindow = JavaSampler.this.lastWindow.getAndUpdate(previous -> Math.max(this.window, previous));
+            if (previousWindow != 0 && previousWindow != this.window) {
+
+                // collect statistics for the previous window
+                JavaSampler.this.windowStatisticsCollector.measureNow(previousWindow);
+
+                // prune data older than the history size
+                IntPredicate predicate = ProfilingWindowUtils.keepHistoryBefore(this.window);
+                JavaSampler.this.dataAggregator.pruneData(predicate);
+                JavaSampler.this.windowStatisticsCollector.pruneStatistics(predicate);
             }
         }
     }
 
     @Override
-    public SamplerData toProto(SparkPlatform platform, CommandSender creator, Comparator<ThreadNode> outputOrder, String comment, MergeMode mergeMode, ClassSourceLookup classSourceLookup) {
+    public SamplerData toProto(SparkPlatform platform, CommandSender creator, String comment, MergeMode mergeMode, ClassSourceLookup classSourceLookup) {
         SamplerData.Builder proto = SamplerData.newBuilder();
         writeMetadataToProto(proto, platform, creator, comment, this.dataAggregator);
-        writeDataToProto(proto, this.dataAggregator, outputOrder, mergeMode, classSourceLookup);
+        writeDataToProto(proto, this.dataAggregator, mergeMode, classSourceLookup);
         return proto.build();
     }
 
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/SimpleDataAggregator.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/SimpleDataAggregator.java
index 39e21aa..54173fe 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/SimpleDataAggregator.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/SimpleDataAggregator.java
@@ -44,8 +44,8 @@ public class SimpleDataAggregator extends JavaDataAggregator {
     }
 
     @Override
-    public void insertData(ThreadInfo threadInfo) {
-        writeData(threadInfo);
+    public void insertData(ThreadInfo threadInfo, int window) {
+        writeData(threadInfo, window);
     }
 
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/TickedDataAggregator.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/TickedDataAggregator.java
index e062f31..d537b96 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/java/TickedDataAggregator.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/java/TickedDataAggregator.java
@@ -23,6 +23,7 @@ package me.lucko.spark.common.sampler.java;
 import me.lucko.spark.common.sampler.ThreadGrouper;
 import me.lucko.spark.common.sampler.aggregator.DataAggregator;
 import me.lucko.spark.common.sampler.node.ThreadNode;
+import me.lucko.spark.common.sampler.window.WindowStatisticsCollector;
 import me.lucko.spark.common.tick.TickHook;
 import me.lucko.spark.proto.SparkSamplerProtos.SamplerMetadata;
 
@@ -31,7 +32,6 @@ import java.util.ArrayList;
 import java.util.List;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.TimeUnit;
-import java.util.concurrent.atomic.AtomicInteger;
 
 /**
  * Implementation of {@link DataAggregator} which supports only including sampling data from "ticks"
@@ -48,14 +48,15 @@ public class TickedDataAggregator extends JavaDataAggregator {
     /** The expected number of samples in each tick */
     private final int expectedSize;
 
-    /** The number of ticks aggregated so far */
-    private final AtomicInteger numberOfTicks = new AtomicInteger();
-
-    private final Object mutex = new Object();
+    /** Counts the number of ticks aggregated */
+    private WindowStatisticsCollector.ExplicitTickCounter tickCounter;
 
     // state
     private int currentTick = -1;
-    private TickList currentData = new TickList(0);
+    private TickList currentData = null;
+
+    // guards currentData
+    private final Object mutex = new Object();
 
     public TickedDataAggregator(ExecutorService workerPool, ThreadGrouper threadGrouper, int interval, boolean ignoreSleeping, boolean ignoreNative, TickHook tickHook, int tickLengthThreshold) {
         super(workerPool, threadGrouper, interval, ignoreSleeping, ignoreNative);
@@ -66,29 +67,34 @@ public class TickedDataAggregator extends JavaDataAggregator {
         this.expectedSize = (int) ((50 / intervalMilliseconds) + 10);
     }
 
+    public void setTickCounter(WindowStatisticsCollector.ExplicitTickCounter tickCounter) {
+        this.tickCounter = tickCounter;
+    }
+
     @Override
     public SamplerMetadata.DataAggregator getMetadata() {
         // push the current tick (so numberOfTicks is accurate)
         synchronized (this.mutex) {
             pushCurrentTick();
+            this.currentData = null;
         }
 
         return SamplerMetadata.DataAggregator.newBuilder()
                 .setType(SamplerMetadata.DataAggregator.Type.TICKED)
                 .setThreadGrouper(this.threadGrouper.asProto())
                 .setTickLengthThreshold(this.tickLengthThreshold)
-                .setNumberOfIncludedTicks(this.numberOfTicks.get())
+                .setNumberOfIncludedTicks(this.tickCounter.getTotalCountedTicks())
                 .build();
     }
 
     @Override
-    public void insertData(ThreadInfo threadInfo) {
+    public void insertData(ThreadInfo threadInfo, int window) {
         synchronized (this.mutex) {
             int tick = this.tickHook.getCurrentTick();
-            if (this.currentTick != tick) {
+            if (this.currentTick != tick || this.currentData == null) {
                 pushCurrentTick();
                 this.currentTick = tick;
-                this.currentData = new TickList(this.expectedSize);
+                this.currentData = new TickList(this.expectedSize, window);
             }
 
             this.currentData.addData(threadInfo);
@@ -98,6 +104,9 @@ public class TickedDataAggregator extends JavaDataAggregator {
     // guarded by 'mutex'
     private void pushCurrentTick() {
         TickList currentData = this.currentData;
+        if (currentData == null) {
+            return;
+        }
 
         // approximate how long the tick lasted
         int tickLengthMicros = currentData.getList().size() * this.interval;
@@ -107,8 +116,8 @@ public class TickedDataAggregator extends JavaDataAggregator {
             return;
         }
 
-        this.numberOfTicks.incrementAndGet();
         this.workerPool.submit(currentData);
+        this.tickCounter.increment();
     }
 
     @Override
@@ -121,21 +130,19 @@ public class TickedDataAggregator extends JavaDataAggregator {
         return super.exportData();
     }
 
-    public int getNumberOfTicks() {
-        return this.numberOfTicks.get();
-    }
-
     private final class TickList implements Runnable {
         private final List<ThreadInfo> list;
+        private final int window;
 
-        TickList(int expectedSize) {
+        TickList(int expectedSize, int window) {
             this.list = new ArrayList<>(expectedSize);
+            this.window = window;
         }
 
         @Override
         public void run() {
             for (ThreadInfo data : this.list) {
-                writeData(data);
+                writeData(data, this.window);
             }
         }
 
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/node/AbstractNode.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/node/AbstractNode.java
index fd2be8d..163365c 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/node/AbstractNode.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/node/AbstractNode.java
@@ -20,69 +20,81 @@
 
 package me.lucko.spark.common.sampler.node;
 
+import me.lucko.spark.common.sampler.window.ProtoTimeEncoder;
+
 import java.util.ArrayList;
 import java.util.Collection;
 import java.util.Collections;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 import java.util.concurrent.ConcurrentHashMap;
 import java.util.concurrent.atomic.LongAdder;
+import java.util.function.IntPredicate;
 
 /**
  * Encapsulates a timed node in the sampling stack.
  */
 public abstract class AbstractNode {
 
-    private static final int MAX_STACK_DEPTH = 300;
+    protected static final int MAX_STACK_DEPTH = Integer.getInteger("spark.maxStackDepth", 300);
 
     /** A map of the nodes children */
     private final Map<StackTraceNode.Description, StackTraceNode> children = new ConcurrentHashMap<>();
 
     /** The accumulated sample time for this node, measured in microseconds */
-    private final LongAdder totalTime = new LongAdder();
+    // Integer key = the window (effectively System.currentTimeMillis() / 60_000)
+    // LongAdder value = accumulated time in microseconds
+    private final Map<Integer, LongAdder> times = new ConcurrentHashMap<>();
 
     /**
-     * Gets the total sample time logged for this node in milliseconds.
+     * Gets the time accumulator for a given window
      *
-     * @return the total time
+     * @param window the window
+     * @return the accumulator
      */
-    public double getTotalTime() {
-        return this.totalTime.longValue() / 1000d;
+    protected LongAdder getTimeAccumulator(int window) {
+        LongAdder adder = this.times.get(window);
+        if (adder == null) {
+            adder = new LongAdder();
+            this.times.put(window, adder);
+        }
+        return adder;
     }
 
-    public Collection<StackTraceNode> getChildren() {
-        return this.children.values();
+    /**
+     * Gets the time windows that have been logged for this node.
+     *
+     * @return the time windows
+     */
+    public Set<Integer> getTimeWindows() {
+        return this.times.keySet();
     }
 
     /**
-     * Logs the given stack trace against this node and its children.
+     * Removes time windows from this node if they pass the given {@code predicate} test.
      *
-     * @param describer the function that describes the elements of the stack
-     * @param stack the stack
-     * @param time the total time to log
-     * @param <T> the stack trace element type
+     * @param predicate the predicate
+     * @return true if any time windows were removed
      */
-    public <T> void log(StackTraceNode.Describer<T> describer, T[] stack, long time) {
-        if (stack.length == 0) {
-            return;
-        }
-
-        this.totalTime.add(time);
-
-        AbstractNode node = this;
-        T previousElement = null;
-
-        for (int offset = 0; offset < Math.min(MAX_STACK_DEPTH, stack.length); offset++) {
-            T element = stack[(stack.length - 1) - offset];
+    public boolean removeTimeWindows(IntPredicate predicate) {
+        return this.times.keySet().removeIf(predicate::test);
+    }
 
-            node = node.resolveChild(describer.describe(element, previousElement));
-            node.totalTime.add(time);
+    /**
+     * Gets the encoded total sample times logged for this node in milliseconds.
+     *
+     * @return the total times
+     */
+    protected double[] encodeTimesForProto(ProtoTimeEncoder encoder) {
+        return encoder.encode(this.times);
+    }
 
-            previousElement = element;
-        }
+    public Collection<StackTraceNode> getChildren() {
+        return this.children.values();
     }
 
-    private StackTraceNode resolveChild(StackTraceNode.Description description) {
+    protected StackTraceNode resolveChild(StackTraceNode.Description description) {
         StackTraceNode result = this.children.get(description); // fast path
         if (result != null) {
             return result;
@@ -96,7 +108,7 @@ public abstract class AbstractNode {
      * @param other the other node
      */
     protected void merge(AbstractNode other) {
-        this.totalTime.add(other.totalTime.longValue());
+        other.times.forEach((key, value) -> getTimeAccumulator(key).add(value.longValue()));
         for (Map.Entry<StackTraceNode.Description, StackTraceNode> child : other.children.entrySet()) {
             resolveChild(child.getKey()).merge(child.getValue());
         }
@@ -123,7 +135,6 @@ public abstract class AbstractNode {
             list.add(child);
         }
 
-        list.sort(null);
         return list;
     }
 
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/node/StackTraceNode.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/node/StackTraceNode.java
index b0d9237..c0dcc5b 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/node/StackTraceNode.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/node/StackTraceNode.java
@@ -20,6 +20,7 @@
 
 package me.lucko.spark.common.sampler.node;
 
+import me.lucko.spark.common.sampler.window.ProtoTimeEncoder;
 import me.lucko.spark.common.util.MethodDisambiguator;
 import me.lucko.spark.proto.SparkSamplerProtos;
 
@@ -30,7 +31,7 @@ import java.util.Objects;
 /**
  * Represents a stack trace element within the {@link AbstractNode node} structure.
  */
-public final class StackTraceNode extends AbstractNode implements Comparable<StackTraceNode> {
+public final class StackTraceNode extends AbstractNode {
 
     /**
      * Magic number to denote "no present" line number for a node.
@@ -64,12 +65,16 @@ public final class StackTraceNode extends AbstractNode implements Comparable<Sta
         return this.description.parentLineNumber;
     }
 
-    public SparkSamplerProtos.StackTraceNode toProto(MergeMode mergeMode) {
+    public SparkSamplerProtos.StackTraceNode toProto(MergeMode mergeMode, ProtoTimeEncoder timeEncoder, Iterable<Integer> childrenRefs) {
         SparkSamplerProtos.StackTraceNode.Builder proto = SparkSamplerProtos.StackTraceNode.newBuilder()
-                .setTime(getTotalTime())
                 .setClassName(this.description.className)
                 .setMethodName(this.description.methodName);
 
+        double[] times = encodeTimesForProto(timeEncoder);
+        for (double time : times) {
+            proto.addTimes(time);
+        }
+
         if (this.description.lineNumber >= 0) {
             proto.setLineNumber(this.description.lineNumber);
         }
@@ -86,27 +91,11 @@ public final class StackTraceNode extends AbstractNode implements Comparable<Sta
                     .ifPresent(proto::setMethodDesc);
         }
 
-        for (StackTraceNode child : exportChildren(mergeMode)) {
-            proto.addChildren(child.toProto(mergeMode));
-        }
+        proto.addAllChildrenRefs(childrenRefs);
 
         return proto.build();
     }
 
-    @Override
-    public int compareTo(StackTraceNode that) {
-        if (this == that) {
-            return 0;
-        }
-
-        int i = -Double.compare(this.getTotalTime(), that.getTotalTime());
-        if (i != 0) {
-            return i;
-        }
-
-        return this.description.compareTo(that.description);
-    }
-
     /**
      * Function to construct a {@link StackTraceNode.Description} from a stack trace element
      * of type {@code T}.
@@ -129,7 +118,7 @@ public final class StackTraceNode extends AbstractNode implements Comparable<Sta
     /**
      * Encapsulates the attributes of a {@link StackTraceNode}.
      */
-    public static final class Description implements Comparable<Description> {
+    public static final class Description {
         private final String className;
         private final String methodName;
 
@@ -162,54 +151,6 @@ public final class StackTraceNode extends AbstractNode implements Comparable<Sta
             this.hash = Objects.hash(this.className, this.methodName, this.methodDescription);
         }
 
-        private static <T extends Comparable<T>> int nullCompare(T a, T b) {
-            if (a == null && b == null) {
-                return 0;
-            } else if (a == null) {
-                return -1;
-            } else if (b == null) {
-                return 1;
-            } else {
-                return a.compareTo(b);
-            }
-        }
-
-        @Override
-        public int compareTo(Description that) {
-            if (this == that) {
-                return 0;
-            }
-
-            int i = this.className.compareTo(that.className);
-            if (i != 0) {
-                return i;
-            }
-
-            i = this.methodName.compareTo(that.methodName);
-            if (i != 0) {
-                return i;
-            }
-
-            i = nullCompare(this.methodDescription, that.methodDescription);
-            if (i != 0) {
-                return i;
-            }
-
-            if (this.methodDescription != null && that.methodDescription != null) {
-                i = this.methodDescription.compareTo(that.methodDescription);
-                if (i != 0) {
-                    return i;
-                }
-            }
-
-            i = Integer.compare(this.lineNumber, that.lineNumber);
-            if (i != 0) {
-                return i;
-            }
-
-            return Integer.compare(this.parentLineNumber, that.parentLineNumber);
-        }
-
         @Override
         public boolean equals(Object o) {
             if (this == o) return true;
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/node/ThreadNode.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/node/ThreadNode.java
index ed97443..37ff359 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/sampler/node/ThreadNode.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/node/ThreadNode.java
@@ -20,8 +20,19 @@
 
 package me.lucko.spark.common.sampler.node;
 
+import me.lucko.spark.common.sampler.window.ProtoTimeEncoder;
+import me.lucko.spark.common.util.IndexedListBuilder;
 import me.lucko.spark.proto.SparkSamplerProtos;
 
+import java.util.ArrayDeque;
+import java.util.Collection;
+import java.util.Deque;
+import java.util.Iterator;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Queue;
+import java.util.function.IntPredicate;
+
 /**
  * The root of a sampling stack for a given thread / thread group.
  */
@@ -53,15 +64,162 @@ public final class ThreadNode extends AbstractNode {
         this.label = label;
     }
 
-    public SparkSamplerProtos.ThreadNode toProto(MergeMode mergeMode) {
+    /**
+     * Logs the given stack trace against this node and its children.
+     *
+     * @param describer the function that describes the elements of the stack
+     * @param stack the stack
+     * @param time the total time to log
+     * @param window the window
+     * @param <T> the stack trace element type
+     */
+    public <T> void log(StackTraceNode.Describer<T> describer, T[] stack, long time, int window) {
+        if (stack.length == 0) {
+            return;
+        }
+
+        getTimeAccumulator(window).add(time);
+
+        AbstractNode node = this;
+        T previousElement = null;
+
+        for (int offset = 0; offset < Math.min(MAX_STACK_DEPTH, stack.length); offset++) {
+            T element = stack[(stack.length - 1) - offset];
+
+            node = node.resolveChild(describer.describe(element, previousElement));
+            node.getTimeAccumulator(window).add(time);
+
+            previousElement = element;
+        }
+    }
+
+    /**
+     * Removes time windows that match the given {@code predicate}.
+     *
+     * @param predicate the predicate to use to test the time windows
+     * @return true if this node is now empty
+     */
+    public boolean removeTimeWindowsRecursively(IntPredicate predicate) {
+        Queue<AbstractNode> queue = new ArrayDeque<>();
+        queue.add(this);
+
+        while (!queue.isEmpty()) {
+            AbstractNode node = queue.remove();
+            Collection<StackTraceNode> children = node.getChildren();
+
+            boolean needToProcessChildren = false;
+
+            for (Iterator<StackTraceNode> it = children.iterator(); it.hasNext(); ) {
+                StackTraceNode child = it.next();
+
+                boolean windowsWereRemoved = child.removeTimeWindows(predicate);
+                boolean childIsNowEmpty = child.getTimeWindows().isEmpty();
+
+                if (childIsNowEmpty) {
+                    it.remove();
+                    continue;
+                }
+
+                if (windowsWereRemoved) {
+                    needToProcessChildren = true;
+                }
+            }
+
+            if (needToProcessChildren) {
+                queue.addAll(children);
+            }
+        }
+
+        removeTimeWindows(predicate);
+        return getTimeWindows().isEmpty();
+    }
+
+    public SparkSamplerProtos.ThreadNode toProto(MergeMode mergeMode, ProtoTimeEncoder timeEncoder) {
         SparkSamplerProtos.ThreadNode.Builder proto = SparkSamplerProtos.ThreadNode.newBuilder()
-                .setName(getThreadLabel())
-                .setTime(getTotalTime());
+                .setName(getThreadLabel());
+
+        double[] times = encodeTimesForProto(timeEncoder);
+        for (double time : times) {
+            proto.addTimes(time);
+        }
+
+        // When converting to a proto, we change the data structure from a recursive tree to an array.
+        // Effectively, instead of:
+        //
+        //   {
+        //     data: 'one',
+        //     children: [
+        //       {
+        //         data: 'two',
+        //         children: [{ data: 'four' }]
+        //       },
+        //       { data: 'three' }
+        //     ]
+        //   }
+        //
+        // we transmit:
+        //
+        //   [
+        //     { data: 'one', children: [1, 2] },
+        //     { data: 'two', children: [3] }
+        //     { data: 'three', children: [] }
+        //     { data: 'four', children: [] }
+        //   ]
+        //
+
+        // the flattened array of nodes
+        IndexedListBuilder<SparkSamplerProtos.StackTraceNode> nodesArray = new IndexedListBuilder<>();
 
+        // Perform a depth-first post order traversal of the tree
+        Deque<Node> stack = new ArrayDeque<>();
+
+        // push the thread node's children to the stack
+        List<Integer> childrenRefs = new LinkedList<>();
         for (StackTraceNode child : exportChildren(mergeMode)) {
-            proto.addChildren(child.toProto(mergeMode));
+            stack.push(new Node(child, childrenRefs));
+        }
+
+        Node node;
+        while (!stack.isEmpty()) {
+            node = stack.peek();
+
+            // on the first visit, just push this node's children and leave it on the stack
+            if (node.firstVisit) {
+                for (StackTraceNode child : node.stackTraceNode.exportChildren(mergeMode)) {
+                    stack.push(new Node(child, node.childrenRefs));
+                }
+                node.firstVisit = false;
+                continue;
+            }
+
+            // convert StackTraceNode to a proto
+            // - at this stage, we have already visited this node's children
+            // - the refs for each child are stored in node.childrenRefs
+            SparkSamplerProtos.StackTraceNode childProto = node.stackTraceNode.toProto(mergeMode, timeEncoder, node.childrenRefs);
+
+            // add the child proto to the nodes array, and record the ref in the parent
+            int childIndex = nodesArray.add(childProto);
+            node.parentChildrenRefs.add(childIndex);
+
+            // pop from the stack
+            stack.pop();
         }
 
+        proto.addAllChildrenRefs(childrenRefs);
+        proto.addAllChildren(nodesArray.build());
+
         return proto.build();
     }
+
+    private static final class Node {
+        private final StackTraceNode stackTraceNode;
+        private boolean firstVisit = true;
+        private final List<Integer> childrenRefs = new LinkedList<>();
+        private final List<Integer> parentChildrenRefs;
+
+        private Node(StackTraceNode node, List<Integer> parentChildrenRefs) {
+            this.stackTraceNode = node;
+            this.parentChildrenRefs = parentChildrenRefs;
+        }
+    }
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/source/ClassSourceLookup.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/source/ClassSourceLookup.java
new file mode 100644
index 0000000..ab63c00
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/source/ClassSourceLookup.java
@@ -0,0 +1,462 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.source;
+
+import me.lucko.spark.common.SparkPlatform;
+import me.lucko.spark.common.sampler.node.StackTraceNode;
+import me.lucko.spark.common.sampler.node.ThreadNode;
+import me.lucko.spark.common.util.ClassFinder;
+
+import org.checkerframework.checker.nullness.qual.Nullable;
+
+import java.io.IOException;
+import java.net.MalformedURLException;
+import java.net.URISyntaxException;
+import java.net.URL;
+import java.net.URLClassLoader;
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.security.CodeSource;
+import java.security.ProtectionDomain;
+import java.util.ArrayDeque;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.Objects;
+import java.util.Queue;
+import java.util.function.Function;
+import java.util.stream.Collectors;
+
+/**
+ * A function which defines the source of given {@link Class}es or (Mixin) method calls.
+ */
+public interface ClassSourceLookup {
+
+    /**
+     * Identify the given class.
+     *
+     * @param clazz the class
+     * @return the source of the class
+     */
+    @Nullable String identify(Class<?> clazz) throws Exception;
+
+    /**
+     * Identify the given method call.
+     *
+     * @param methodCall the method call info
+     * @return the source of the method call
+     */
+    default @Nullable String identify(MethodCall methodCall) throws Exception {
+        return null;
+    }
+
+    /**
+     * Identify the given method call.
+     *
+     * @param methodCall the method call info
+     * @return the source of the method call
+     */
+    default @Nullable String identify(MethodCallByLine methodCall) throws Exception {
+        return null;
+    }
+
+    /**
+     * A no-operation {@link ClassSourceLookup}.
+     */
+    ClassSourceLookup NO_OP = new ClassSourceLookup() {
+        @Override
+        public @Nullable String identify(Class<?> clazz) {
+            return null;
+        }
+    };
+
+    static ClassSourceLookup create(SparkPlatform platform) {
+        try {
+            return platform.createClassSourceLookup();
+        } catch (Exception e) {
+            e.printStackTrace();
+            return NO_OP;
+        }
+    }
+
+    /**
+     * A {@link ClassSourceLookup} which identifies classes based on their {@link ClassLoader}.
+     */
+    abstract class ByClassLoader implements ClassSourceLookup {
+
+        public abstract @Nullable String identify(ClassLoader loader) throws Exception;
+
+        @Override
+        public final @Nullable String identify(Class<?> clazz) throws Exception {
+            ClassLoader loader = clazz.getClassLoader();
+            while (loader != null) {
+                String source = identify(loader);
+                if (source != null) {
+                    return source;
+                }
+                loader = loader.getParent();
+            }
+            return null;
+        }
+    }
+
+    /**
+     * A {@link ClassSourceLookup} which identifies classes based on URL.
+     */
+    interface ByUrl extends ClassSourceLookup {
+
+        default String identifyUrl(URL url) throws URISyntaxException, MalformedURLException {
+            Path path = null;
+
+            String protocol = url.getProtocol();
+            if (protocol.equals("file")) {
+                path = Paths.get(url.toURI());
+            } else if (protocol.equals("jar")) {
+                URL innerUrl = new URL(url.getPath());
+                path = Paths.get(innerUrl.getPath().split("!")[0]);
+            }
+
+            if (path != null) {
+                return identifyFile(path.toAbsolutePath().normalize());
+            }
+
+            return null;
+        }
+
+        default String identifyFile(Path path) {
+            return identifyFileName(path.getFileName().toString());
+        }
+
+        default String identifyFileName(String fileName) {
+            return fileName.endsWith(".jar") ? fileName.substring(0, fileName.length() - 4) : null;
+        }
+    }
+
+    /**
+     * A {@link ClassSourceLookup} which identifies classes based on the first URL in a {@link URLClassLoader}.
+     */
+    class ByFirstUrlSource extends ClassSourceLookup.ByClassLoader implements ClassSourceLookup.ByUrl {
+        @Override
+        public @Nullable String identify(ClassLoader loader) throws IOException, URISyntaxException {
+            if (loader instanceof URLClassLoader) {
+                URLClassLoader urlClassLoader = (URLClassLoader) loader;
+                URL[] urls = urlClassLoader.getURLs();
+                if (urls.length == 0) {
+                    return null;
+                }
+                return identifyUrl(urls[0]);
+            }
+            return null;
+        }
+    }
+
+    /**
+     * A {@link ClassSourceLookup} which identifies classes based on their {@link ProtectionDomain#getCodeSource()}.
+     */
+    class ByCodeSource implements ClassSourceLookup, ClassSourceLookup.ByUrl {
+        @Override
+        public @Nullable String identify(Class<?> clazz) throws URISyntaxException, MalformedURLException {
+            ProtectionDomain protectionDomain = clazz.getProtectionDomain();
+            if (protectionDomain == null) {
+                return null;
+            }
+            CodeSource codeSource = protectionDomain.getCodeSource();
+            if (codeSource == null) {
+                return null;
+            }
+
+            URL url = codeSource.getLocation();
+            return url == null ? null : identifyUrl(url);
+        }
+    }
+
+    interface Visitor {
+        void visit(ThreadNode node);
+
+        boolean hasClassSourceMappings();
+
+        Map<String, String> getClassSourceMapping();
+
+        boolean hasMethodSourceMappings();
+
+        Map<String, String> getMethodSourceMapping();
+
+        boolean hasLineSourceMappings();
+
+        Map<String, String> getLineSourceMapping();
+    }
+
+    static Visitor createVisitor(ClassSourceLookup lookup) {
+        if (lookup == ClassSourceLookup.NO_OP) {
+            return NoOpVisitor.INSTANCE; // don't bother!
+        }
+        return new VisitorImpl(lookup);
+    }
+
+    enum NoOpVisitor implements Visitor {
+        INSTANCE;
+
+        @Override
+        public void visit(ThreadNode node) {
+
+        }
+
+        @Override
+        public boolean hasClassSourceMappings() {
+            return false;
+        }
+
+        @Override
+        public Map<String, String> getClassSourceMapping() {
+            return Collections.emptyMap();
+        }
+
+        @Override
+        public boolean hasMethodSourceMappings() {
+            return false;
+        }
+
+        @Override
+        public Map<String, String> getMethodSourceMapping() {
+            return Collections.emptyMap();
+        }
+
+        @Override
+        public boolean hasLineSourceMappings() {
+            return false;
+        }
+
+        @Override
+        public Map<String, String> getLineSourceMapping() {
+            return Collections.emptyMap();
+        }
+    }
+
+    /**
+     * Visitor which scans {@link StackTraceNode}s and accumulates class/method call identities.
+     */
+    class VisitorImpl implements Visitor {
+        private final ClassSourceLookup lookup;
+        private final ClassFinder classFinder = new ClassFinder();
+
+        private final SourcesMap<String> classSources = new SourcesMap<>(Function.identity());
+        private final SourcesMap<MethodCall> methodSources = new SourcesMap<>(MethodCall::toString);
+        private final SourcesMap<MethodCallByLine> lineSources = new SourcesMap<>(MethodCallByLine::toString);
+
+        VisitorImpl(ClassSourceLookup lookup) {
+            this.lookup = lookup;
+        }
+
+        @Override
+        public void visit(ThreadNode node) {
+            Queue<StackTraceNode> queue = new ArrayDeque<>(node.getChildren());
+            for (StackTraceNode n = queue.poll(); n != null; n = queue.poll()) {
+                visitStackNode(n);
+                queue.addAll(n.getChildren());
+            }
+        }
+
+        private void visitStackNode(StackTraceNode node) {
+            this.classSources.computeIfAbsent(
+                    node.getClassName(),
+                    className -> {
+                        Class<?> clazz = this.classFinder.findClass(className);
+                        if (clazz == null) {
+                            return null;
+                        }
+                        return this.lookup.identify(clazz);
+                    });
+
+            if (node.getMethodDescription() != null) {
+                MethodCall methodCall = new MethodCall(node.getClassName(), node.getMethodName(), node.getMethodDescription());
+                this.methodSources.computeIfAbsent(methodCall, this.lookup::identify);
+            } else {
+                MethodCallByLine methodCall = new MethodCallByLine(node.getClassName(), node.getMethodName(), node.getLineNumber());
+                this.lineSources.computeIfAbsent(methodCall, this.lookup::identify);
+            }
+        }
+
+        @Override
+        public boolean hasClassSourceMappings() {
+            return this.classSources.hasMappings();
+        }
+
+        @Override
+        public Map<String, String> getClassSourceMapping() {
+            return this.classSources.export();
+        }
+
+        @Override
+        public boolean hasMethodSourceMappings() {
+            return this.methodSources.hasMappings();
+        }
+
+        @Override
+        public Map<String, String> getMethodSourceMapping() {
+            return this.methodSources.export();
+        }
+
+        @Override
+        public boolean hasLineSourceMappings() {
+            return this.lineSources.hasMappings();
+        }
+
+        @Override
+        public Map<String, String> getLineSourceMapping() {
+            return this.lineSources.export();
+        }
+    }
+
+    final class SourcesMap<T> {
+        // <key> --> identifier (plugin name)
+        private final Map<T, String> map = new HashMap<>();
+        private final Function<? super T, String> keyToStringFunction;
+
+        private SourcesMap(Function<? super T, String> keyToStringFunction) {
+            this.keyToStringFunction = keyToStringFunction;
+        }
+
+        public void computeIfAbsent(T key, ComputeSourceFunction<T> function) {
+            if (!this.map.containsKey(key)) {
+                try {
+                    this.map.put(key, function.compute(key));
+                } catch (Throwable e) {
+                    this.map.put(key, null);
+                }
+            }
+        }
+
+        public boolean hasMappings() {
+            this.map.values().removeIf(Objects::isNull);
+            return !this.map.isEmpty();
+        }
+
+        public Map<String, String> export() {
+            this.map.values().removeIf(Objects::isNull);
+            if (this.keyToStringFunction.equals(Function.identity())) {
+                //noinspection unchecked
+                return (Map<String, String>) this.map;
+            } else {
+                return this.map.entrySet().stream().collect(Collectors.toMap(
+                        e -> this.keyToStringFunction.apply(e.getKey()),
+                        Map.Entry::getValue
+                ));
+            }
+        }
+
+        private interface ComputeSourceFunction<T> {
+            String compute(T key) throws Exception;
+        }
+    }
+
+    /**
+     * Encapsulates information about a given method call using the name + method description.
+     */
+    final class MethodCall {
+        private final String className;
+        private final String methodName;
+        private final String methodDescriptor;
+
+        public MethodCall(String className, String methodName, String methodDescriptor) {
+            this.className = className;
+            this.methodName = methodName;
+            this.methodDescriptor = methodDescriptor;
+        }
+
+        public String getClassName() {
+            return this.className;
+        }
+
+        public String getMethodName() {
+            return this.methodName;
+        }
+
+        public String getMethodDescriptor() {
+            return this.methodDescriptor;
+        }
+
+        @Override
+        public String toString() {
+            return this.className + ";" + this.methodName + ";" + this.methodDescriptor;
+        }
+
+        @Override
+        public boolean equals(Object o) {
+            if (this == o) return true;
+            if (!(o instanceof MethodCall)) return false;
+            MethodCall that = (MethodCall) o;
+            return this.className.equals(that.className) &&
+                    this.methodName.equals(that.methodName) &&
+                    this.methodDescriptor.equals(that.methodDescriptor);
+        }
+
+        @Override
+        public int hashCode() {
+            return Objects.hash(this.className, this.methodName, this.methodDescriptor);
+        }
+    }
+
+    /**
+     * Encapsulates information about a given method call using the name + line number.
+     */
+    final class MethodCallByLine {
+        private final String className;
+        private final String methodName;
+        private final int lineNumber;
+
+        public MethodCallByLine(String className, String methodName, int lineNumber) {
+            this.className = className;
+            this.methodName = methodName;
+            this.lineNumber = lineNumber;
+        }
+
+        public String getClassName() {
+            return this.className;
+        }
+
+        public String getMethodName() {
+            return this.methodName;
+        }
+
+        public int getLineNumber() {
+            return this.lineNumber;
+        }
+
+        @Override
+        public String toString() {
+            return this.className + ";" + this.lineNumber;
+        }
+
+        @Override
+        public boolean equals(Object o) {
+            if (this == o) return true;
+            if (!(o instanceof MethodCallByLine)) return false;
+            MethodCallByLine that = (MethodCallByLine) o;
+            return this.lineNumber == that.lineNumber && this.className.equals(that.className);
+        }
+
+        @Override
+        public int hashCode() {
+            return Objects.hash(this.className, this.lineNumber);
+        }
+    }
+
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/source/SourceMetadata.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/source/SourceMetadata.java
new file mode 100644
index 0000000..0808d66
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/source/SourceMetadata.java
@@ -0,0 +1,81 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.source;
+
+import com.google.common.collect.ImmutableList;
+
+import me.lucko.spark.proto.SparkSamplerProtos.SamplerMetadata;
+
+import java.util.Collection;
+import java.util.List;
+import java.util.function.Function;
+
+/**
+ * A "source" is a plugin or mod on the platform that may be identified
+ * as a source of a method call in a profile.
+ */
+public class SourceMetadata {
+
+    public static <T> List<SourceMetadata> gather(Collection<T> sources, Function<? super T, String> nameFunction, Function<? super T, String> versionFunction, Function<? super T, String> authorFunction) {
+        ImmutableList.Builder<SourceMetadata> builder = ImmutableList.builder();
+
+        for (T source : sources) {
+            String name = nameFunction.apply(source);
+            String version = versionFunction.apply(source);
+            String author = authorFunction.apply(source);
+
+            SourceMetadata metadata = new SourceMetadata(name, version, author);
+            builder.add(metadata);
+        }
+
+        return builder.build();
+    }
+
+    private final String name;
+    private final String version;
+    private final String author;
+
+    public SourceMetadata(String name, String version, String author) {
+        this.name = name;
+        this.version = version;
+        this.author = author;
+    }
+
+    public String getName() {
+        return this.name;
+    }
+
+    public String getVersion() {
+        return this.version;
+    }
+
+    public String getAuthor() {
+        return this.author;
+    }
+
+    public SamplerMetadata.SourceMetadata toProto() {
+        return SamplerMetadata.SourceMetadata.newBuilder()
+                .setName(this.name)
+                .setVersion(this.version)
+                .build();
+    }
+
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/window/ProfilingWindowUtils.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/window/ProfilingWindowUtils.java
new file mode 100644
index 0000000..be6f08a
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/window/ProfilingWindowUtils.java
@@ -0,0 +1,70 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.window;
+
+import me.lucko.spark.common.sampler.aggregator.DataAggregator;
+
+import java.util.function.IntPredicate;
+
+public enum ProfilingWindowUtils {
+    ;
+
+    /**
+     * The size/duration of a profiling window in seconds.
+     * (1 window = 1 minute)
+     */
+    public static final int WINDOW_SIZE_SECONDS = 60;
+
+    /**
+     * The number of windows to record in continuous profiling before data is dropped.
+     * (60 windows * 1 minute = 1 hour of profiling data)
+     */
+    public static final int HISTORY_SIZE = Integer.getInteger("spark.continuousProfilingHistorySize", 60);
+
+    /**
+     * Gets the profiling window for the given time in unix-millis.
+     *
+     * @param time the time in milliseconds
+     * @return the window
+     */
+    public static int unixMillisToWindow(long time) {
+        return (int) (time / (WINDOW_SIZE_SECONDS * 1000L));
+    }
+
+    /**
+     * Gets the window at the current time.
+     *
+     * @return the window
+     */
+    public static int windowNow() {
+        return unixMillisToWindow(System.currentTimeMillis());
+    }
+
+    /**
+     * Gets a prune predicate that can be passed to {@link DataAggregator#pruneData(IntPredicate)}.
+     *
+     * @return the prune predicate
+     */
+    public static IntPredicate keepHistoryBefore(int currentWindow) {
+        // windows that were earlier than (currentWindow minus history size) should be pruned
+        return window -> window < (currentWindow - HISTORY_SIZE);
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/window/ProtoTimeEncoder.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/window/ProtoTimeEncoder.java
new file mode 100644
index 0000000..03da075
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/window/ProtoTimeEncoder.java
@@ -0,0 +1,93 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.window;
+
+import me.lucko.spark.common.sampler.async.jfr.Dictionary;
+import me.lucko.spark.common.sampler.node.ThreadNode;
+
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.atomic.LongAdder;
+import java.util.stream.IntStream;
+
+/**
+ * Encodes a map of int->double into a double array.
+ */
+public class ProtoTimeEncoder {
+    /** A sorted array of all possible keys to encode */
+    private final int[] keys;
+    /** A map of key value -> index in the keys array */
+    private final Map<Integer, Integer> keysToIndex;
+
+    public ProtoTimeEncoder(List<ThreadNode> sourceData) {
+        // get an array of all keys that show up in the source data
+        this.keys = sourceData.stream()
+                .map(n -> n.getTimeWindows().stream().mapToInt(i -> i))
+                .reduce(IntStream.empty(), IntStream::concat)
+                .distinct()
+                .sorted()
+                .toArray();
+
+        // construct a reverse index lookup
+        this.keysToIndex = new HashMap<>(this.keys.length);
+        for (int i = 0; i < this.keys.length; i++) {
+            this.keysToIndex.put(this.keys[i], i);
+        }
+    }
+
+    /**
+     * Gets an array of the keys that could be encoded by this encoder.
+     *
+     * @return an array of keys
+     */
+    public int[] getKeys() {
+        return this.keys;
+    }
+
+    /**
+     * Encode a {@link Dictionary} (map) of times/durations into a double array.
+     *
+     * @param times a dictionary of times (unix-time millis -> duration in microseconds)
+     * @return the times encoded as a double array
+     */
+    public double[] encode(Map<Integer, LongAdder> times) {
+        // construct an array of values - length needs to exactly match the
+        // number of keys, even if some values are zero.
+        double[] array = new double[this.keys.length];
+
+        times.forEach((key, value) -> {
+            // get the index for the given key
+            Integer idx = this.keysToIndex.get(key);
+            if (idx == null) {
+                throw new RuntimeException("No index for key " + key + " in " + this.keysToIndex.keySet());
+            }
+
+            // convert the duration from microseconds -> milliseconds
+            double durationInMilliseconds = value.longValue() / 1000d;
+
+            // store in the array
+            array[idx] = durationInMilliseconds;
+        });
+
+        return array;
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/sampler/window/WindowStatisticsCollector.java b/spark-common/src/main/java/me/lucko/spark/common/sampler/window/WindowStatisticsCollector.java
new file mode 100644
index 0000000..ce65013
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/sampler/window/WindowStatisticsCollector.java
@@ -0,0 +1,287 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.sampler.window;
+
+import me.lucko.spark.common.SparkPlatform;
+import me.lucko.spark.common.monitor.cpu.CpuMonitor;
+import me.lucko.spark.common.monitor.tick.TickStatistics;
+import me.lucko.spark.common.platform.world.AsyncWorldInfoProvider;
+import me.lucko.spark.common.platform.world.WorldInfoProvider;
+import me.lucko.spark.common.tick.TickHook;
+import me.lucko.spark.common.util.RollingAverage;
+import me.lucko.spark.proto.SparkProtos;
+
+import java.util.Map;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.atomic.AtomicInteger;
+import java.util.function.IntPredicate;
+
+/**
+ * Collects statistics for each profiling window.
+ */
+public class WindowStatisticsCollector {
+    private static final SparkProtos.WindowStatistics ZERO = SparkProtos.WindowStatistics.newBuilder().build();
+
+    /** The platform */
+    private final SparkPlatform platform;
+
+    /** Map of profiling window -> statistics */
+    private final Map<Integer, SparkProtos.WindowStatistics> stats;
+
+    private TickCounter tickCounter;
+
+    public WindowStatisticsCollector(SparkPlatform platform) {
+        this.platform = platform;
+        this.stats = new ConcurrentHashMap<>();
+    }
+
+    /**
+     * Indicates to the statistics collector that it should count the number
+     * of ticks in each window using the provided {@link TickHook}.
+     *
+     * @param hook the tick hook
+     */
+    public void startCountingTicks(TickHook hook) {
+        this.tickCounter = new NormalTickCounter(this.platform, hook);
+    }
+
+    /**
+     * Indicates to the statistics collector that it should count the number
+     * of ticks in each window, according to how many times the
+     * {@link ExplicitTickCounter#increment()} method is called.
+     *
+     * @param hook the tick hook
+     * @return the counter
+     */
+    public ExplicitTickCounter startCountingTicksExplicit(TickHook hook) {
+        ExplicitTickCounter counter = new ExplicitTickCounter(this.platform, hook);
+        this.tickCounter = counter;
+        return counter;
+    }
+
+    public void stop() {
+        if (this.tickCounter != null) {
+            this.tickCounter.stop();
+        }
+    }
+
+    /**
+     * Gets the total number of ticks that have passed between the time
+     * when the profiler started and stopped.
+     *
+     * <p>Importantly, note that this metric is different to the total number of ticks in a window
+     * (which is recorded by {@link SparkProtos.WindowStatistics#getTicks()}) or the total number
+     * of observed ticks if the 'only-ticks-over' aggregator is being used
+     * (which is recorded by {@link SparkProtos.WindowStatistics#getTicks()}
+     * and {@link ExplicitTickCounter#getTotalCountedTicks()}.</p>
+     *
+     * @return the total number of ticks in the profile
+     */
+    public int getTotalTicks() {
+        return this.tickCounter == null ? -1 : this.tickCounter.getTotalTicks();
+    }
+
+    /**
+     * Measures statistics for the given window if none have been recorded yet.
+     *
+     * @param window the window
+     */
+    public void measureNow(int window) {
+        this.stats.computeIfAbsent(window, w -> measure());
+    }
+
+    /**
+     * Ensures that the exported map has statistics (even if they are zeroed) for all windows.
+     *
+     * @param windows the expected windows
+     */
+    public void ensureHasStatisticsForAllWindows(int[] windows) {
+        for (int window : windows) {
+            this.stats.computeIfAbsent(window, w -> ZERO);
+        }
+    }
+
+    public void pruneStatistics(IntPredicate predicate) {
+        this.stats.keySet().removeIf(predicate::test);
+    }
+
+    public Map<Integer, SparkProtos.WindowStatistics> export() {
+        return this.stats;
+    }
+
+    /**
+     * Measures current statistics, where possible averaging over the last minute. (1 min = 1 window)
+     *
+     * @return the current statistics
+     */
+    private SparkProtos.WindowStatistics measure() {
+        SparkProtos.WindowStatistics.Builder builder = SparkProtos.WindowStatistics.newBuilder();
+
+        TickStatistics tickStatistics = this.platform.getTickStatistics();
+        if (tickStatistics != null) {
+            builder.setTps(tickStatistics.tps1Min());
+
+            RollingAverage mspt = tickStatistics.duration1Min();
+            if (mspt != null) {
+                builder.setMsptMedian(mspt.median());
+                builder.setMsptMax(mspt.max());
+            }
+        }
+
+        if (this.tickCounter != null) {
+            int ticks = this.tickCounter.getCountedTicksThisWindowAndReset();
+            builder.setTicks(ticks);
+        }
+
+        builder.setCpuProcess(CpuMonitor.processLoad1MinAvg());
+        builder.setCpuSystem(CpuMonitor.systemLoad1MinAvg());
+
+        try {
+            AsyncWorldInfoProvider worldInfoProvider = new AsyncWorldInfoProvider(this.platform, this.platform.getPlugin().createWorldInfoProvider());
+            WorldInfoProvider.CountsResult counts = worldInfoProvider.getCounts();
+            if (counts != null) {
+                builder.setPlayers(counts.players());
+                builder.setEntities(counts.entities());
+                builder.setTileEntities(counts.tileEntities());
+                builder.setChunks(counts.chunks());
+            }
+        } catch (Exception e) {
+            e.printStackTrace();
+        }
+
+        return builder.build();
+    }
+
+    /**
+     * Responsible for counting the number of ticks in a profile/window.
+     */
+    public interface TickCounter {
+
+        /**
+         * Stop the counter.
+         */
+        void stop();
+
+        /**
+         * Get the total number of ticks.
+         *
+         * <p>See {@link WindowStatisticsCollector#getTotalTicks()} for a longer explanation
+         * of what this means exactly.</p>
+         *
+         * @return the total ticks
+         */
+        int getTotalTicks();
+
+        /**
+         * Gets the total number of ticks counted in the last window,
+         * and resets the counter to zero.
+         *
+         * @return the number of ticks counted since the last time this method was called
+         */
+        int getCountedTicksThisWindowAndReset();
+    }
+
+    private static abstract class BaseTickCounter implements TickCounter {
+        protected final SparkPlatform platform;
+        protected final TickHook tickHook;
+
+        /** The game tick when sampling first began */
+        private final int startTick;
+
+        /** The game tick when sampling stopped */
+        private int stopTick = -1;
+
+        BaseTickCounter(SparkPlatform platform, TickHook tickHook) {
+            this.platform = platform;
+            this.tickHook = tickHook;
+            this.startTick = this.tickHook.getCurrentTick();
+        }
+
+        @Override
+        public void stop() {
+            this.stopTick = this.tickHook.getCurrentTick();
+        }
+
+        @Override
+        public int getTotalTicks() {
+            if (this.startTick == -1) {
+                throw new IllegalStateException("start tick not recorded");
+            }
+            if (this.stopTick == -1) {
+                throw new IllegalStateException("stop tick not recorded");
+            }
+
+            return this.stopTick - this.startTick;
+        }
+    }
+
+    /**
+     * Counts the number of ticks in a window using a {@link TickHook}.
+     */
+    public static final class NormalTickCounter extends BaseTickCounter {
+        private int last;
+
+        NormalTickCounter(SparkPlatform platform, TickHook tickHook) {
+            super(platform, tickHook);
+            this.last = this.tickHook.getCurrentTick();
+        }
+
+        @Override
+        public int getCountedTicksThisWindowAndReset() {
+            synchronized (this) {
+                int now = this.tickHook.getCurrentTick();
+                int ticks = now - this.last;
+                this.last = now;
+                return ticks;
+            }
+        }
+    }
+
+    /**
+     * Counts the number of ticks in a window according to the number of times
+     * {@link #increment()} is called.
+     *
+     * Used by the {@link me.lucko.spark.common.sampler.java.TickedDataAggregator}.
+     */
+    public static final class ExplicitTickCounter extends BaseTickCounter {
+        private final AtomicInteger counted = new AtomicInteger();
+        private final AtomicInteger total = new AtomicInteger();
+
+        ExplicitTickCounter(SparkPlatform platform, TickHook tickHook) {
+            super(platform, tickHook);
+        }
+
+        public void increment() {
+            this.counted.incrementAndGet();
+            this.total.incrementAndGet();
+        }
+
+        public int getTotalCountedTicks() {
+            return this.total.get();
+        }
+
+        @Override
+        public int getCountedTicksThisWindowAndReset() {
+            return this.counted.getAndSet(0);
+        }
+    }
+
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/util/ClassSourceLookup.java b/spark-common/src/main/java/me/lucko/spark/common/util/ClassSourceLookup.java
deleted file mode 100644
index bd9ec37..0000000
--- a/spark-common/src/main/java/me/lucko/spark/common/util/ClassSourceLookup.java
+++ /dev/null
@@ -1,241 +0,0 @@
-/*
- * This file is part of spark.
- *
- *  Copyright (c) lucko (Luck) <luck@lucko.me>
- *  Copyright (c) contributors
- *
- *  This program is free software: you can redistribute it and/or modify
- *  it under the terms of the GNU General Public License as published by
- *  the Free Software Foundation, either version 3 of the License, or
- *  (at your option) any later version.
- *
- *  This program is distributed in the hope that it will be useful,
- *  but WITHOUT ANY WARRANTY; without even the implied warranty of
- *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *  GNU General Public License for more details.
- *
- *  You should have received a copy of the GNU General Public License
- *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
- */
-
-package me.lucko.spark.common.util;
-
-import me.lucko.spark.common.sampler.node.StackTraceNode;
-import me.lucko.spark.common.sampler.node.ThreadNode;
-
-import org.checkerframework.checker.nullness.qual.Nullable;
-
-import java.io.IOException;
-import java.net.MalformedURLException;
-import java.net.URISyntaxException;
-import java.net.URL;
-import java.net.URLClassLoader;
-import java.nio.file.Path;
-import java.nio.file.Paths;
-import java.security.CodeSource;
-import java.security.ProtectionDomain;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.Map;
-import java.util.Objects;
-
-/**
- * A function which defines the source of given {@link Class}es.
- */
-public interface ClassSourceLookup {
-
-    /**
-     * Identify the given class.
-     *
-     * @param clazz the class
-     * @return the source of the class
-     */
-    @Nullable String identify(Class<?> clazz) throws Exception;
-
-    /**
-     * A no-operation {@link ClassSourceLookup}.
-     */
-    ClassSourceLookup NO_OP = new ClassSourceLookup() {
-        @Override
-        public @Nullable String identify(Class<?> clazz) {
-            return null;
-        }
-    };
-
-    /**
-     * A {@link ClassSourceLookup} which identifies classes based on their {@link ClassLoader}.
-     */
-    abstract class ByClassLoader implements ClassSourceLookup {
-
-        public abstract @Nullable String identify(ClassLoader loader) throws Exception;
-
-        @Override
-        public final @Nullable String identify(Class<?> clazz) throws Exception {
-            ClassLoader loader = clazz.getClassLoader();
-            while (loader != null) {
-                String source = identify(loader);
-                if (source != null) {
-                    return source;
-                }
-                loader = loader.getParent();
-            }
-            return null;
-        }
-    }
-
-    /**
-     * A {@link ClassSourceLookup} which identifies classes based on URL.
-     */
-    interface ByUrl extends ClassSourceLookup {
-
-        default String identifyUrl(URL url) throws URISyntaxException, MalformedURLException {
-            Path path = null;
-
-            String protocol = url.getProtocol();
-            if (protocol.equals("file")) {
-                path = Paths.get(url.toURI());
-            } else if (protocol.equals("jar")) {
-                URL innerUrl = new URL(url.getPath());
-                path = Paths.get(innerUrl.getPath().split("!")[0]);
-            }
-
-            if (path != null) {
-                return identifyFile(path.toAbsolutePath().normalize());
-            }
-
-            return null;
-        }
-
-        default String identifyFile(Path path) {
-            return identifyFileName(path.getFileName().toString());
-        }
-
-        default String identifyFileName(String fileName) {
-            return fileName.endsWith(".jar") ? fileName.substring(0, fileName.length() - 4) : null;
-        }
-    }
-
-    /**
-     * A {@link ClassSourceLookup} which identifies classes based on the first URL in a {@link URLClassLoader}.
-     */
-    class ByFirstUrlSource extends ByClassLoader implements ByUrl {
-        @Override
-        public @Nullable String identify(ClassLoader loader) throws IOException, URISyntaxException {
-            if (loader instanceof URLClassLoader) {
-                URLClassLoader urlClassLoader = (URLClassLoader) loader;
-                URL[] urls = urlClassLoader.getURLs();
-                if (urls.length == 0) {
-                    return null;
-                }
-                return identifyUrl(urls[0]);
-            }
-            return null;
-        }
-    }
-
-    /**
-     * A {@link ClassSourceLookup} which identifies classes based on their {@link ProtectionDomain#getCodeSource()}.
-     */
-    class ByCodeSource implements ClassSourceLookup, ByUrl {
-        @Override
-        public @Nullable String identify(Class<?> clazz) throws URISyntaxException, MalformedURLException {
-            ProtectionDomain protectionDomain = clazz.getProtectionDomain();
-            if (protectionDomain == null) {
-                return null;
-            }
-            CodeSource codeSource = protectionDomain.getCodeSource();
-            if (codeSource == null) {
-                return null;
-            }
-
-            URL url = codeSource.getLocation();
-            return url == null ? null : identifyUrl(url);
-        }
-    }
-
-    interface Visitor {
-        void visit(ThreadNode node);
-
-        boolean hasMappings();
-
-        Map<String, String> getMapping();
-    }
-
-    static Visitor createVisitor(ClassSourceLookup lookup) {
-        if (lookup == ClassSourceLookup.NO_OP) {
-            return NoOpVisitor.INSTANCE; // don't bother!
-        }
-        return new VisitorImpl(lookup);
-    }
-
-    enum NoOpVisitor implements Visitor {
-        INSTANCE;
-
-        @Override
-        public void visit(ThreadNode node) {
-
-        }
-
-        @Override
-        public boolean hasMappings() {
-            return false;
-        }
-
-        @Override
-        public Map<String, String> getMapping() {
-            return Collections.emptyMap();
-        }
-    }
-
-    /**
-     * Visitor which scans {@link StackTraceNode}s and accumulates class identities.
-     */
-    class VisitorImpl implements Visitor {
-        private final ClassSourceLookup lookup;
-        private final ClassFinder classFinder = new ClassFinder();
-
-        // class name --> identifier (plugin name)
-        private final Map<String, String> map = new HashMap<>();
-
-        VisitorImpl(ClassSourceLookup lookup) {
-            this.lookup = lookup;
-        }
-
-        @Override
-        public void visit(ThreadNode node) {
-            for (StackTraceNode child : node.getChildren()) {
-                visitStackNode(child);
-            }
-        }
-
-        @Override
-        public boolean hasMappings() {
-            return !this.map.isEmpty();
-        }
-
-        @Override
-        public Map<String, String> getMapping() {
-            this.map.values().removeIf(Objects::isNull);
-            return this.map;
-        }
-
-        private void visitStackNode(StackTraceNode node) {
-            String className = node.getClassName();
-            if (!this.map.containsKey(className)) {
-                try {
-                    Class<?> clazz = this.classFinder.findClass(className);
-                    Objects.requireNonNull(clazz);
-                    this.map.put(className, this.lookup.identify(clazz));
-                } catch (Throwable e) {
-                    this.map.put(className, null);
-                }
-            }
-
-            // recursively
-            for (StackTraceNode child : node.getChildren()) {
-                visitStackNode(child);
-            }
-        }
-    }
-
-}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/util/Configuration.java b/spark-common/src/main/java/me/lucko/spark/common/util/Configuration.java
index 7588645..32f3bc6 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/util/Configuration.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/util/Configuration.java
@@ -20,32 +20,58 @@
 
 package me.lucko.spark.common.util;
 
+import com.google.gson.Gson;
+import com.google.gson.GsonBuilder;
 import com.google.gson.JsonElement;
 import com.google.gson.JsonObject;
-import com.google.gson.JsonParser;
 import com.google.gson.JsonPrimitive;
 
 import java.io.BufferedReader;
+import java.io.BufferedWriter;
 import java.io.IOException;
 import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
 
 public final class Configuration {
-    private static final JsonParser PARSER = new JsonParser();
+    private static final Gson GSON = new GsonBuilder().setPrettyPrinting().create();
 
-    private final JsonObject root;
+    private final Path file;
+    private JsonObject root;
 
     public Configuration(Path file) {
+        this.file = file;
+        load();
+    }
+
+    public void load() {
         JsonObject root = null;
-        if (Files.exists(file)) {
-            try (BufferedReader reader = Files.newBufferedReader(file, StandardCharsets.UTF_8)) {
-                root = PARSER.parse(reader).getAsJsonObject();
+        if (Files.exists(this.file)) {
+            try (BufferedReader reader = Files.newBufferedReader(this.file, StandardCharsets.UTF_8)) {
+                root = GSON.fromJson(reader, JsonObject.class);
             } catch (IOException e) {
                 e.printStackTrace();
             }
         }
-        this.root = root != null ? root : new JsonObject();
+        if (root == null) {
+            root = new JsonObject();
+            root.addProperty("_header", "spark configuration file - https://spark.lucko.me/docs/Configuration");
+        }
+        this.root = root;
+    }
+
+    public void save() {
+        try {
+            Files.createDirectories(this.file.getParent());
+        } catch (IOException e) {
+            // ignore
+        }
+
+        try (BufferedWriter writer = Files.newBufferedWriter(this.file, StandardCharsets.UTF_8)) {
+            GSON.toJson(this.root, writer);
+        } catch (IOException e) {
+            e.printStackTrace();
+        }
     }
 
     public String getString(String path, String def) {
@@ -67,4 +93,34 @@ public final class Configuration {
         return val.isBoolean() ? val.getAsBoolean() : def;
     }
 
+    public int getInteger(String path, int def) {
+        JsonElement el = this.root.get(path);
+        if (el == null || !el.isJsonPrimitive()) {
+            return def;
+        }
+
+        JsonPrimitive val = el.getAsJsonPrimitive();
+        return val.isBoolean() ? val.getAsInt() : def;
+    }
+
+    public void setString(String path, String value) {
+        this.root.add(path, new JsonPrimitive(value));
+    }
+
+    public void setBoolean(String path, boolean value) {
+        this.root.add(path, new JsonPrimitive(value));
+    }
+
+    public void setInteger(String path, int value) {
+        this.root.add(path, new JsonPrimitive(value));
+    }
+
+    public boolean contains(String path) {
+        return this.root.has(path);
+    }
+
+    public void remove(String path) {
+        this.root.remove(path);
+    }
+
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/util/FormatUtil.java b/spark-common/src/main/java/me/lucko/spark/common/util/FormatUtil.java
index c4a3d66..1ee3b0f 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/util/FormatUtil.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/util/FormatUtil.java
@@ -62,4 +62,24 @@ public enum FormatUtil {
                 .append(Component.text(unit))
                 .build();
     }
+
+    public static String formatSeconds(long seconds) {
+        if (seconds <= 0) {
+            return "0s";
+        }
+
+        long second = seconds;
+        long minute = second / 60;
+        second = second % 60;
+
+        StringBuilder sb = new StringBuilder();
+        if (minute != 0) {
+            sb.append(minute).append("m ");
+        }
+        if (second != 0) {
+            sb.append(second).append("s ");
+        }
+
+        return sb.toString().trim();
+    }
 }
diff --git a/spark-common/src/main/java/me/lucko/spark/common/util/IndexedListBuilder.java b/spark-common/src/main/java/me/lucko/spark/common/util/IndexedListBuilder.java
new file mode 100644
index 0000000..b2315f9
--- /dev/null
+++ b/spark-common/src/main/java/me/lucko/spark/common/util/IndexedListBuilder.java
@@ -0,0 +1,43 @@
+/*
+ * This file is part of spark.
+ *
+ *  Copyright (c) lucko (Luck) <luck@lucko.me>
+ *  Copyright (c) contributors
+ *
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package me.lucko.spark.common.util;
+
+import java.util.ArrayList;
+import java.util.List;
+
+/**
+ * List builder that returns the index of the inserted element.
+ *
+ * @param <T> generic type
+ */
+public class IndexedListBuilder<T> {
+    private int i = 0;
+    private final List<T> nodes = new ArrayList<>();
+
+    public int add(T node) {
+        this.nodes.add(node);
+        return this.i++;
+    }
+
+    public List<T> build() {
+        return this.nodes;
+    }
+}
diff --git a/spark-common/src/main/java/me/lucko/spark/common/util/TemporaryFiles.java b/spark-common/src/main/java/me/lucko/spark/common/util/TemporaryFiles.java
index 8a4a621..91a474c 100644
--- a/spark-common/src/main/java/me/lucko/spark/common/util/TemporaryFiles.java
+++ b/spark-common/src/main/java/me/lucko/spark/common/util/TemporaryFiles.java
@@ -20,10 +20,18 @@
 
 package me.lucko.spark.common.util;
 
+import com.google.common.collect.ImmutableList;
+
 import java.io.IOException;
+import java.nio.charset.StandardCharsets;
+import java.nio.file.FileSystems;
 import java.nio.file.Files;
 import java.nio.file.Path;
+import java.nio.file.attribute.FileAttribute;
+import java.nio.file.attribute.PosixFilePermission;
+import java.nio.file.attribute.PosixFilePermissions;
 import java.util.Collections;
+import java.util.EnumSet;
 import java.util.HashSet;
 import java.util.Iterator;
 import java.util.Set;
@@ -32,23 +40,47 @@ import java.util.Set;
  * Utility for handling temporary files.
  */
 public final class TemporaryFiles {
-    private TemporaryFiles() {}
 
-    private static final Set<Path> DELETE_SET = Collections.synchronizedSet(new HashSet<>());
+    public static final FileAttribute<?>[] OWNER_ONLY_FILE_PERMISSIONS;
+
+    static {
+        boolean isPosix = FileSystems.getDefault().supportedFileAttributeViews().contains("posix");
+        if (isPosix) {
+            OWNER_ONLY_FILE_PERMISSIONS = new FileAttribute[]{PosixFilePermissions.asFileAttribute(EnumSet.of(
+                    PosixFilePermission.OWNER_READ, PosixFilePermission.OWNER_WRITE
+            ))};
+        } else {
+            OWNER_ONLY_FILE_PERMISSIONS = new FileAttribute[0];
+        }
+    }
+
+    private final Path tmpDirectory;
+    private final Set<Path> files = Collections.synchronizedSet(new HashSet<>());
 
-    public static Path create(String prefix, String suffix) throws IOException {
-        return register(Files.createTempFile(prefix, suffix));
+    public TemporaryFiles(Path tmpDirectory) {
+        this.tmpDirectory = tmpDirectory;
     }
 
-    public static Path register(Path path) {
+    public Path create(String prefix, String suffix) throws IOException {
+        Path file;
+        if (ensureDirectoryIsReady()) {
+            String name = prefix + Long.toHexString(System.nanoTime()) + suffix;
+            file = Files.createFile(this.tmpDirectory.resolve(name), OWNER_ONLY_FILE_PERMISSIONS);
+        } else {
+            file = Files.createTempFile(prefix, suffix);
+        }
+        return register(file);
+    }
+
+    public Path register(Path path) {
         path.toFile().deleteOnExit();
-        DELETE_SET.add(path);
+        this.files.add(path);
         return path;
     }
 
-    public static void deleteTemporaryFiles() {
-        synchronized (DELETE_SET) {
-            for (Iterator<Path> iterator = DELETE_SET.iterator(); iterator.hasNext(); ) {
+    public void deleteTemporaryFiles() {
+        synchronized (this.files) {
+            for (Iterator<Path> iterator = this.files.iterator(); iterator.hasNext(); ) {
                 Path path = iterator.next();
                 try {
                     Files.deleteIfExists(path);
@@ -60,4 +92,35 @@ public final class TemporaryFiles {
         }
     }
 
+    private boolean ensureDirectoryIsReady() {
+        if (Boolean.parseBoolean(System.getProperty("spark.useOsTmpDir", "false"))) {
+            return false;
+        }
+
+        if (Files.isDirectory(this.tmpDirectory)) {
+            return true;
+        }
+
+        try {
+            Files.createDirectories(this.tmpDirectory);
+
+            Files.write(this.tmpDirectory.resolve("about.txt"), ImmutableList.of(
+                    "# What is this directory?",
+                    "",
+                    "* In order to perform certain functions, spark sometimes needs to write temporary data to the disk. ",
+                    "* Previously, a temporary directory provided by the operating system was used for this purpose. ",
+                    "* However, this proved to be unreliable in some circumstances, so spark now stores temporary data here instead!",
+                    "",
+                    "spark will automatically cleanup the contents of this directory. " ,
+                    "(but if for some reason it doesn't, if the server is stopped, you can freely delete any files ending in .tmp)",
+                    "",
+                    "tl;dr: spark uses this folder to store some temporary data."
+            ), StandardCharsets.UTF_8);
+
+            return true;
+        } catch (IOException e) {
+            return false;
+        }
+    }
+
 }
diff --git a/spark-common/src/main/proto/spark/spark.proto b/spark-common/src/main/proto/spark/spark.proto
index 2ea341f..f61e585 100644
--- a/spark-common/src/main/proto/spark/spark.proto
+++ b/spark-common/src/main/proto/spark/spark.proto
@@ -152,6 +152,21 @@ message WorldStatistics {
   }
 }
 
+message WindowStatistics {
+  int32 ticks = 1;
+  double cpu_process = 2;
+  double cpu_system = 3;
+  double tps = 4;
+  double mspt_median = 5;
+  double mspt_max = 6;
+
+  // world
+  int32 players = 7;
+  int32 entities = 8;
+  int32 tile_entities = 9;
+  int32 chunks = 10;
+}
+
 message RollingAverageValues {
   double mean = 1;
   double max = 2;
diff --git a/spark-common/src/main/proto/spark/spark_sampler.proto b/spark-common/src/main/proto/spark/spark_sampler.proto
index 8d9512a..245da37 100644
--- a/spark-common/src/main/proto/spark/spark_sampler.proto
+++ b/spark-common/src/main/proto/spark/spark_sampler.proto
@@ -11,6 +11,10 @@ message SamplerData {
   SamplerMetadata metadata = 1;
   repeated ThreadNode threads = 2;
   map<string, string> class_sources = 3; // optional
+  map<string, string> method_sources = 4; // optional
+  map<string, string> line_sources = 5; // optional
+  repeated int32 time_windows = 6;
+  map<int32, WindowStatistics> time_window_statistics = 7;
 }
 
 message SamplerMetadata {
@@ -26,6 +30,8 @@ message SamplerMetadata {
   map<string, string> server_configurations = 10;
   int64 end_time = 11;
   int32 number_of_ticks = 12;
+  map<string, SourceMetadata> sources = 13;
+  map<string, string> extra_platform_metadata = 14;
 
   message ThreadDumper {
     Type type = 1;
@@ -56,20 +62,35 @@ message SamplerMetadata {
       AS_ONE = 2;
     }
   }
+
+  message SourceMetadata {
+    string name = 1;
+    string version = 2;
+  }
 }
 
 message ThreadNode {
   string name = 1;
-  double time = 2;
+
+  // replaced
+  reserved 2;
+  reserved "time";
+
   repeated StackTraceNode children = 3;
+  repeated double times = 4;
+  repeated int32 children_refs = 5;
 }
 
 message StackTraceNode {
-  double time = 1;
-  repeated StackTraceNode children = 2;
+  // replaced
+  reserved 1, 2;
+  reserved "time", "children";
+
   string class_name = 3;
   string method_name = 4;
   int32 parent_line_number = 5; // optional
   int32 line_number = 6; // optional
   string method_desc = 7; // optional
+  repeated double times = 8;
+  repeated int32 children_refs = 9;
 }
diff --git a/spark-common/src/main/resources/spark/linux/aarch64/libasyncProfiler.so b/spark-common/src/main/resources/spark/linux/aarch64/libasyncProfiler.so
index cf6c48b..800cf91 100755
--- a/spark-common/src/main/resources/spark/linux/aarch64/libasyncProfiler.so
+++ b/spark-common/src/main/resources/spark/linux/aarch64/libasyncProfiler.so
diff --git a/spark-common/src/main/resources/spark/linux/amd64-musl/libasyncProfiler.so b/spark-common/src/main/resources/spark/linux/amd64-musl/libasyncProfiler.so
index 0a08f7c..3c81d1c 100755
--- a/spark-common/src/main/resources/spark/linux/amd64-musl/libasyncProfiler.so
+++ b/spark-common/src/main/resources/spark/linux/amd64-musl/libasyncProfiler.so
diff --git a/spark-common/src/main/resources/spark/linux/amd64/libasyncProfiler.so b/spark-common/src/main/resources/spark/linux/amd64/libasyncProfiler.so
index 0deb9e0..5af5071 100755
--- a/spark-common/src/main/resources/spark/linux/amd64/libasyncProfiler.so
+++ b/spark-common/src/main/resources/spark/linux/amd64/libasyncProfiler.so
diff --git a/spark-common/src/main/resources/spark/macos/libasyncProfiler.so b/spark-common/src/main/resources/spark/macos/libasyncProfiler.so
index 65b4aed..4930c67 100755
--- a/spark-common/src/main/resources/spark/macos/libasyncProfiler.so
+++ b/spark-common/src/main/resources/spark/macos/libasyncProfiler.so