A gossip protocol spreads information through a cluster by letting nodes talk to a few other nodes over and over again.

Instead of one node sending an update to every other node, each node shares what it knows with a few peers. Those peers do the same in later rounds. After enough rounds, the update reaches most or all of the cluster.

This works well because information has many paths through the system. If one message is dropped, another message may carry the same update later. If one node is down, the update can still spread through other nodes.

Gossip is useful for information that does not need to be perfect immediately, such as:

Gossip complements consensus. Gossip tells nodes about state over time. Consensus makes a group agree on one decision before moving forward.

This chapter covers how gossip spreads state, what it can and cannot promise, and where teams use it in real systems.