longpolling

Long polling is a web communication technique in which a client requests data from a server and the server holds the request open until new data becomes available or a timeout occurs. When data is available, the server responds with the update. The client then immediately issues a new request, creating a near-continuous stream of updates without maintaining a persistent, full-duplex connection.

How long polling works: The client sends an HTTP request to a server endpoint. If there is no new data, the server waits for a specified period. If new data arrives before the timeout, the server responds with the data. If the timeout expires first, the server responds with an empty result, and the client issues a new request. Upon receiving data, the client processes it and repeats the cycle. This pattern can be implemented with standard HTTP/HTTPS and works with common server environments and reverse proxies.

Compared to simple polling, long polling reduces latency by delivering data as soon as it is available rather than at fixed intervals. Compared to WebSockets or server-sent events, long polling can be easier to deploy in environments with strict proxy or load-balancer configurations, since it uses standard HTTP requests. However, long polling can be more resource-intensive for servers, as each connected client holds a request open for some time, and it may require careful scaling with event-driven servers, connection pooling, and, in some cases, sticky sessions.

Typical use cases include chat, live notifications, and real-time feeds. Security considerations include using HTTPS, authenticating clients, and validating data. Limitations include potential latency spikes and higher resource usage under large numbers of concurrent connections.