What is TCP keep-alive and when might it be used?

• A. It is a one-time check when a connection is established.
• B. Periodic probes to verify an idle connection is alive; used for long-lived idle sessions, often configurable.
• C. It increases the window size automatically.
• D. It terminates idle connections after a fixed time.

Answer: (B)

TCP keep-alive is a mechanism to check that an idle connection is still reachable by sending periodic probes after a period of silence. If the other side responds, the connection remains open; if there’s no response after a configured number of probes, the connection can be closed, which helps detect broken paths or crashed peers in long-lived connections. It’s especially useful for long-lived idle sessions where data isn’t constantly flowing, such as persistent client-server ties that might pass through NATs or firewalls that time out idle state. The behavior is configurable: you set how long to wait before the first probe, how often probes are sent, and how many failed probes trigger termination. It doesn’t apply to the initial handshake, doesn’t automatically change window size, and isn’t simply a fixed-time termination mechanism. Keep-alives can add some extra network traffic, so they’re tuned or disabled when not needed.