C10K

C10k refers to the challenge of managing ten thousand concurrent TCP connections on a single server process or machine with acceptable latency and resource usage. Originating in the late 1990s, the term became a focal point in discussions about scalable network services and web servers.

The problem was popularized by Dan Kegel, who described the limitations of conventional server designs that spawn a new thread per connection and rely on blocking I/O. Such designs suffer from high memory usage, context-switching overhead, and difficulties in meeting latency targets at scale.

Typical responses shift away from one-thread-per-connection to event-driven or asynchronous I/O models. This includes non-blocking I/O with multiplexing, where a single thread handles many connections, using facilities such as select, poll, and more scalable interfaces like epoll on Linux, kqueue on BSD/macOS, or IOCP on Windows. Frameworks and servers such as Nginx, Node.js, libuv, and Netty demonstrate practical implementations of these ideas.

Over time, the focus has extended beyond ten thousand to higher scales, but C10k remains a reference point for evaluating the effectiveness of I/O models, thread management, and OS support. It influenced the design of modern high-performance servers and the adoption of asynchronous programming in many ecosystems.