HTTPpersistent

HTTPpersistent refers to the practice of keeping a TCP connection open and reusing it for multiple HTTP requests and responses within a single session. By avoiding the need to close and reopen a connection for each request, HTTPpersistent reduces the latency and CPU overhead associated with connection setup and teardown. In HTTP/1.1, connections are designed to be persistent by default, though either party can signal that a connection should be closed after a given response.

Mechanism: After a response is delivered, the connection is kept alive for potential subsequent requests. The server may impose an idle timeout or a maximum number of requests on a given connection, after which it is closed. Clients and intermediaries may also enforce limits. While the technique improves performance for pages that load multiple resources, it does not eliminate all bottlenecks (e.g., bandwidth, server processing).

Relation to other protocols: HTTPpersistent is distinct from newer protocols such as HTTP/2 and HTTP/3, which provide multiplexed streams over a single connection and thus reduce head-of-line blocking and further improve efficiency. Nevertheless, persistent connections remain in use for backward compatibility and in environments where negotiation of newer protocols is not possible.

History and notes: The concept emerged with early HTTP implementations and became a standard feature in HTTP/1.1, encouraging persistent connections as the default behavior. The term HTTPpersistent is sometimes used informally or in documentation to denote the general habit of persistent HTTP connections rather than a specific software product. See also: HTTP keep-alive, HTTP/2, HTTP/3, connection pooling.