Quorum Consensus

What is Consensus & Quorum?

Achieving consensus in a distributed system means getting multiple independent servers (nodes) to agree on a specific value or state, even if some nodes fail or messages are delayed. This is crucial for reliability and consistency.

Waiting for *all* nodes to respond before proceeding is often impractical due to potential failures. A Quorum is a strategy to overcome this. Instead of requiring agreement from all nodes, a system only needs a response or agreement from a majority subset of the nodes to consider an operation successful or a decision valid.

The Quorum Size

For a cluster of N nodes, a simple majority quorum (Q) is typically calculated as: Q = floor(N / 2) + 1

This ensures that any two successful operations (each requiring a quorum) must have at least one node in common, which helps maintain consistency (this principle is key in algorithms like Paxos and Raft).

Benefits

• Fault Tolerance: The system can continue to operate even if some nodes (up to N - Q) fail or are unreachable.
• Availability: Operations don't have to wait for the slowest or failed nodes.

Use Cases

Quorum principles are fundamental in:

• Distributed Locking: Ensuring only one client holds a lock by acquiring it from a quorum of lock servers.
• Distributed Databases/Storage: Ensuring data consistency during reads and writes by requiring operations to succeed on a quorum of replicas (e.g., DynamoDB's NWR settings).
• Leader Election: Algorithms like Raft and Paxos use quorum voting to elect a leader reliably.
• Commit Protocols: Confirming transaction commits across nodes.

(This visualization demonstrates the basic concept of requiring a majority acknowledgement, rather than a specific complex algorithm like Raft or Paxos.)