Quorums and Majority Voting
What you'll learn: How quorum-based approaches ensure consistency by requiring any two quorums to overlap in at least one node.
The Core Idea
In distributed systems, we can't always rely on contacting every node (some might be down or unreachable). A quorum is the minimum number of nodes that must agree before an operation succeeds. The magic happens when we choose quorum sizes cleverly: any two quorums must overlap in at least one node.
Think of it like a jury system where you need a majority to convict. If 5 out of 9 jurors vote guilty, and later another case needs 5 out of the same 9 jurors, those two groups must share at least one juror. That shared juror carries information between decisions.
How Majority Voting Works
The most common approach uses majority quorums. In a system with N nodes:
- Write quorum (W): Must get acknowledgment from W nodes
- Read quorum (R): Must read from R nodes
- Rule: W + R > N
This guarantees overlap. With 5 nodes, if W=3 and R=3, any write touches at least 3 nodes, and any read touches at least 3 nodes. Since 3+3=6 > 5, they must overlap by at least one node. That overlapping node ensures the read sees the most recent write (or can detect conflicts).
Why This Matters
The overlap property gives us consistency without requiring all nodes to agree. The shared node acts as a "witness" ensuring reads can detect recent writes. This is more available than requiring all nodes (system works even with failures) while still maintaining consistency guarantees.
Key Takeaway: Quorum systems achieve consistency by ensuring any two quorums overlap—typically using majority voting where W + R > N guarantees at least one shared node carries information between operations.