Quorum Systems¶

By requiring a subset of nodes to agree (a quorum) rather than all nodes, systems can tolerate failures while maintaining consistency - any two quorums must overlap, ensuring operations see each other's effects.

Tension Addressed¶

Consistency vs. Availability

Recognizable As¶

Majority voting (N/2+1); read/write quorums (R+W>N); flexible quorum configurations; witness nodes; quorum intersection requirements

Canonical Sources¶

Gifford, "Weighted Voting for Replicated Data" (1979); Thomas, "A Majority Consensus Approach to Concurrency Control" (1979)

Transfers To¶

Distributed databases, consensus protocols, distributed file systems, replicated state machines

The key insight is that quorums don't require all nodes - they require overlapping sets. This allows tuning for read-heavy vs. write-heavy workloads (small read quorum, large write quorum, or vice versa) while maintaining consistency.