Server-Timing Headers — Debug Slow Pages Without Guessing

Updated April 2026

Your API takes 800ms. Is it the database? The cache miss? The template render? Without Server-Timing headers, you are guessing. With them, the browser DevTools show you exactly where the time went.

The Server-Timing header format

# Single metric
Server-Timing: db;dur=45.2

# Multiple metrics
Server-Timing: db;dur=45.2, cache;dur=2.1, render;dur=12.8

# With description
Server-Timing: db;desc="User query";dur=45.2, cache;desc="Redis lookup";dur=2.1

Add to your backend

Express / Node.js

app.get("/api/dashboard", async (req, res) => { const timings = []; // Measure database query const dbStart = performance.now(); const user = await db.query("SELECT * FROM users WHERE id = ?", [req.userId]); timings.push(`db;desc="User query";dur=${(performance.now() - dbStart).toFixed(1)}`); // Measure cache lookup const cacheStart = performance.now(); const cached = await redis.get(`dashboard:${req.userId}`); timings.push(`cache;desc="Redis";dur=${(performance.now() - cacheStart).toFixed(1)}`); // Measure external API const apiStart = performance.now(); const metrics = await fetchMetrics(req.userId); timings.push(`metrics_api;desc="Metrics service";dur=${(performance.now() - apiStart).toFixed(1)}`); res.setHeader("Server-Timing", timings.join(", ")); res.json({ user, metrics });
});

FastAPI / Python

import time
from fastapi import Response

@app.get("/api/dashboard")
async def dashboard(response: Response, user_id: int): timings = [] t = time.perf_counter() user = await db.fetch_user(user_id) timings.append(f'db;desc="User query";dur={((time.perf_counter() - t) * 1000):.1f}') t = time.perf_counter() cached = await redis.get(f"dashboard:{user_id}") timings.append(f'cache;desc="Redis";dur={((time.perf_counter() - t) * 1000):.1f}') response.headers["Server-Timing"] = ", ".join(timings) return {"user": user}

Django middleware

import time

class ServerTimingMiddleware: def __init__(self, get_response): self.get_response = get_response def __call__(self, request): start = time.perf_counter() response = self.get_response(request) duration = (time.perf_counter() - start) * 1000 response["Server-Timing"] = f'total;dur={duration:.1f}' return response

View in Chrome DevTools

Open DevTools → Network → click any request → Timing tab. Your Server-Timing metrics appear at the bottom as a bar chart. Each metric shows as a labelled segment.

Add a reusable timing utility

// utils/timing.js — reuse across routes
class ServerTimer { constructor() { this.metrics = []; } start(name, desc) { return { name, desc, start: performance.now() }; } end(timer) { const dur = (performance.now() - timer.start).toFixed(1); this.metrics.push( timer.desc ? `${timer.name};desc="${timer.desc}";dur=${dur}` : `${timer.name};dur=${dur}` ); } header() { return this.metrics.join(", "); }
}

// Usage
const t = new ServerTimer();
const dbTimer = t.start("db", "User query");
const user = await db.getUser(id);
t.end(dbTimer);

res.setHeader("Server-Timing", t.header());