NOTE
3.4 Gossip
How gossip protocols spread information through randomized peer-to-peer exchanges, why they scale well, and the redundancy and convergence trade-offs they introduce.
This is a historical learning note and may contain outdated or incomplete understanding.
1. What Is Gossip?
A gossip protocol spreads information through repeated peer-to-peer exchanges. Each node periodically communicates with one or more peers, and information propagates through the cluster in a way similar to an epidemic.
Gossip-style protocols are widely used for membership, failure detection, metadata propagation, and eventually disseminating state across large clusters.
2. Basic Workflow
- Each node periodically selects one or more peers, often randomly.
- The peers exchange state, summaries, or updates.
- A node that learns new information can pass it to other peers in later rounds.
- After enough rounds, the information is expected to reach most or all healthy nodes.
3. Advantages
- decentralized operation;
- no single coordinator for ordinary dissemination;
- good scalability for large clusters;
- natural tolerance of transient message loss because information is repeatedly retransmitted through different paths.
4. Trade-offs
4.1 Redundant Messages
Randomized spreading intentionally creates redundancy. A node may receive the same information multiple times, consuming extra network and processing resources.
4.2 Convergence Delay
Information does not arrive everywhere at once. During propagation, different nodes can temporarily hold different views of the system.
Gossip therefore should not be confused with a consensus protocol such as Raft or Paxos. Gossip is primarily a dissemination mechanism; additional rules are needed when the system requires a single authoritative decision.