limitconn

Limitconn is a mechanism used by web servers, reverse proxies, and load balancers to cap the number of simultaneous connections allowed from a client or to a resource. The purpose is to protect against abuse, reduce resource contention, and maintain service availability during traffic spikes or slow client behavior. The exact name and syntax vary by server; many implementations use a directive or module with similar functionality, sometimes called limit_conn or limitconn, and in some setups via modules such as mod_limitipconn in Apache.

In nginx, limit_conn is a core directive that works with a shared memory zone defined by limit_conn_zone and keyed by an identifier such as the client IP. A typical configuration defines limit_conn_zone $binary_remote_addr zone=addr:10m; and then uses limit_conn addr 5 to restrict each client to five simultaneous connections. The feature affects connections regardless of keep-alive status, and when the limit is reached, additional attempts to open new connections are delayed or rejected with an appropriate response.

Limit connections can be applied per server, per location, or per key, allowing granular control over resource usage. Administrators should consider the impact of proxies, NAT, and IPv6 when choosing the key for limits, as these factors can affect the accuracy of per-client enforcement.

Other servers implement similar protections through modules or directives designed to limit per-client concurrency. For example, Apache offers modules such as mod_limitipconn to restrict the number of concurrent connections per IP. Some reverse proxies and load balancers also provide per-client connection caps or related rate-limiting features as part of their traffic control options.

See also: rate limiting, per-user quotas, keep-alive management, connection throttling.