Try it in 30 seconds, without changing your code
npm install traceperf
node --require traceperf/auto server.js
Every request becomes a call tree with its HTTP calls and database queries nested inside. Stop the server and you get per-route statistics and an explanation of what made requests slow:
└─ GET /users/:id: 84.112 ms
├─ pg SELECT users: 12.331 ms
└─ fetch GET billing.internal/invoices: 70.950 ms
🔥 renderPdf ran 30 ms of synchronous code, blocking everything else in the process for that long
⏳ GET /users/:id waited ~60 ms while renderPdf in POST /export blocked the event loop
⇉ GET /dashboard awaited 3 calls one after another; Promise.all could save up to ~47 ms
🔁 GET /orders called pg SELECT prices 10 times; batching them into one call may help (N+1 pattern)
Features
🌳 Call trees across await
Tracked calls inside tracked calls nest automatically, even across concurrent async requests. Total, self and synchronous time for every call.
💡 Explains causes
traceperf explain finds blocked event loops (and the function that blocked), sequential awaits and N+1 queries.
🔌 Zero-code coverage
HTTP servers, fetch, pg, mysql2, MongoDB/Mongoose, ioredis, Prisma, AWS SDK. Express, Fastify, Hono, Koa, NestJS.
🔥 CPU hotspots for slow calls
profileSlowCalls samples the CPU and names the functions and lines that were busy during slow calls.
📈 Production statistics
p50/p95/p99 per label with bounded memory, sampling and a ~45 ns kill switch. Prometheus endpoint included.
🚦 Regressions in CI
traceperf compare before.jsonl after.jsonl --fail fails the build when something got slower.
📤 Gets data out
Chrome/Perfetto traces, OpenTelemetry spans, JSON lines, HTML flame-chart reports, a live terminal view and Chrome DevTools tracks.
🌐 Runs everywhere
Node.js 14+, Bun, Deno, Cloudflare Workers, Vercel Edge and browsers. CommonJS, ESM and TypeScript. No dependencies.
Getting Started
Installation
npm install traceperf
Basic Usage
const { createTracePerf, TrackingMode } = require('traceperf');
// Create a tracker with default settings
const traceperf = createTracePerf();
// Track a synchronous function
const result = traceperf.track(() => {
// Your function logic here
return 'result';
}, { label: 'myFunction' });
// Track an asynchronous function
async function main() {
const result = await traceperf.track(async () => {
// Your async function logic here
return 'async result';
}, { label: 'asyncFunction' });
}
main();
Tracking Modes
const { createTracePerf, TrackingMode } = require('traceperf');
// Performance mode - minimal overhead for production
const performanceTracker = createTracePerf({
trackingMode: TrackingMode.PERFORMANCE,
silent: true,
threshold: 100, // only track functions that take more than 100ms
sampleRate: 0.1 // only track 10% of function calls
});
// Detailed mode - maximum information for development
const devTracker = createTracePerf({
trackingMode: TrackingMode.DETAILED,
trackMemory: true,
enableNestedTracking: true
});
Memory Tracking
const { createTracePerf } = require('traceperf');
// Enable memory tracking
const tracker = createTracePerf({ trackMemory: true });
// Track memory usage for a memory-intensive operation
const result = tracker.track(() => {
// Memory intensive operations
const largeArray = new Array(1000000).fill(Math.random());
return processArray(largeArray);
}, { label: 'memoryIntensiveOperation' });
Output:
└─ memoryIntensiveOperation: 25.005 ms 📊 +15335.69 KB
Nested Function Tracking
const { createTracePerf } = require('traceperf');
const traceperf = createTracePerf({ enableNestedTracking: true });
// Define some functions (call the *tracked* versions so they nest)
function processData(data) {
return trackedTransformData(data);
}
function transformData(data) {
return data.map(x => x * 2);
}
// Create trackable versions
const trackedProcessData = traceperf.createTrackable(processData, {
label: 'processData'
});
const trackedTransformData = traceperf.createTrackable(transformData, {
label: 'transformData'
});
// Execute with tracking
traceperf.track(() => {
const data = [1, 2, 3, 4, 5];
return trackedProcessData(data);
}, { label: 'mainOperation' });
Output:
└─ mainOperation: 0.580 ms 📊 +3.12 KB
└─ processData: 0.330 ms 📊 +1.40 KB
└─ transformData: 0.120 ms 📊 +0.75 KB
Browser Support
<script src="https://unpkg.com/traceperf/dist/traceperf.browser.min.js"></script>
<script>
const { createTracePerf, BrowserLogger } = TracePerf;
const browserTracker = createTracePerf({
logger: new BrowserLogger({
silent: false,
trackMemory: true
})
});
// Track a DOM operation
browserTracker.track(() => {
document.getElementById('output').textContent = 'Updated';
}, { label: 'updateDOM' });
// Track async operations
await browserTracker.track(async () => {
const response = await fetch('/api/data');
return response.json();
}, { label: 'fetchData' });
// Create trackable functions
const trackedFn = browserTracker.createTrackable(() => {
// Function implementation
}, { label: 'trackedBrowserFn' });
</script>
ES Modules Usage
import { createTracePerf, BrowserLogger } from 'traceperf/browser';
const browserTracker = createTracePerf({
logger: new BrowserLogger({
silent: false,
trackMemory: true
})
});
// Use the same API as in Node.js
browserTracker.track(() => {
// Your browser-side code
});
Documentation
For complete documentation and examples, please visit the GitHub repository.
- Running traceperf in production: safe settings, measured costs, rollout
- Compared with clinic.js, --cpu-prof, OpenTelemetry and APM agents
- Overhead benchmarks
- API reference
View examples at examples directory.