How do you optimize Nginx performance for high traffic?

Expected Interview Answer

Step-by-Step Explanation

1. Step 1

   Match workers to CPU

   Set worker_processes auto and raise worker_connections plus the OS ulimit for file descriptors.

2. Step 2

   Reuse connections

   Enable keepalive for clients and add an upstream keepalive pool to avoid repeated TCP and TLS handshakes.

3. Step 3

   Cache aggressively

   Configure proxy_cache or fastcgi_cache so repeat requests are served without hitting the backend.

4. Step 4

   Compress and offload static

   Enable gzip or brotli, use sendfile and tcp_nopush, and serve static files directly.

5. Step 5

   Tune TLS and HTTP/2

   Enable ssl_session_cache, OCSP stapling and HTTP/2 to cut per-connection overhead.

6. Step 6

   Load test and iterate

   Benchmark with tools like wrk, then adjust buffers, timeouts and keepalive values from real data.

Code Example

worker_processes auto;
worker_rlimit_nofile 100000;

events {
    worker_connections 4096;
    multi_accept on;
}

http {
    sendfile on;
    tcp_nopush on;
    keepalive_timeout 65;
    gzip on;

    proxy_cache_path /var/cache/nginx keys_zone=app:100m inactive=60m;

    upstream backend {
        server 127.0.0.1:8080;
        keepalive 64;
    }

    server {
        location / {
            proxy_cache app;
            proxy_pass http://backend;
            proxy_http_version 1.1;
            proxy_set_header Connection "";
        }
    }
}

Follow-up Questions

• How does upstream keepalive differ from client keepalive?
• When should you use proxy_cache versus a dedicated CDN?
• How do you size worker_connections relative to file descriptors?
• What metrics tell you Nginx is the bottleneck versus the backend?
• How does HTTP/2 change connection tuning?