Keep-alive and connection reuse

Persistent HTTP connections amortize TCP and TLS handshakes across multiple requests. Node's `http.Agent` pools sockets with `keepAlive: true` by default in modern versions — critical for services that call many upstream APIs.

					import { Agent, request } from 'node:https';

const agent = new Agent({ keepAlive: true, maxSockets: 50, timeout: 30_000 });

request('https://api.example.com/data', { agent }, (res) => { /* ... */ });
				

Idle timeouts on load balancers, NAT gateways, and upstream servers can close connections silently — the next request may fail and retry on a fresh socket. Socket exhaustion happens when `maxSockets` is too low or connections leak.

On interviews: explain why connection reuse improves P99 latency; align idle timeouts client < LB < server; diagnose `ECONNRESET` after idle periods.

Common pitfalls: creating a new Agent per request; mismatched keep-alive between proxy and app; not handling half-open connections during deploy.

The trade-off is balancing simplicity, performance, safety, and operability — name which axis you optimized and what cost you accepted.

Checklist:

• Enable pooled agents for outbound HTTP.
• Tune maxSockets to concurrency needs.
• Align timeouts across hops.
• Monitor open socket counts under load.