Simple definition
HTTP keep-alive reuses a network connection for multiple requests instead of opening a connection each time.
Purpose and use
HTTP keep-alive avoids rebuilding a transport connection for every exchange eligible for reuse.
How does it work?
HTTP/1.1 connections are persistent by default unless closure is indicated; reuse timeouts and limits depend on clients, servers and proxies.
Example and verification
For failures after an idle period, compare client, reverse-proxy and application-server timeouts.
Points to watch
Connection or Keep-Alive fields intended for HTTP/1.x should not be blindly carried into HTTP/2 or HTTP/3, which have their own rules.
In short
Reuse reduces selected setup costs but guarantees neither a permanent connection nor freedom from idle timeouts.