Performance Debugging: Step by Step

Performance Debugging: Step by Step¶

The application is slow. Where to start? A systematic step-by-step guide.

1. Define the Problem¶

• Which endpoint is slow?
• Current vs target latency?
• Consistent or intermittent?

2. Measure¶

curl -o /dev/null -s -w “%{time_total}\n” URL

3. Identify the Bottleneck¶

• Network — DNS, TLS, TTFB
• Backend — CPU, memory, I/O
• Database — slow queries
• Frontend — render blocking
• External services

4. Backend Profiling¶

node –inspect app.js python -m cProfile app.py go tool pprof …

5. Database¶

EXPLAIN (ANALYZE, BUFFERS) SELECT …;

6. Optimize and Verify¶

One change at a time. Measure before and after.

7. Monitoring¶

• P95/P99 latency
• Query time
• Error rate
• Resource utilization

Summary¶

Measure -> Identify bottleneck -> Optimize -> Verify -> Monitor.