Quorum Trade-offs and Failure Scenarios
You'll learn: How choosing different quorum configurations affects your system's behavior when nodes fail or networks partition.
The Core Idea
Every quorum configuration (N, R, W) represents a deliberate trade-off between consistency, availability, and performance. When failures occur—nodes crash, networks partition, or latency spikes—your quorum settings determine whether operations succeed or fail, and whether you risk serving stale data or losing availability.
Understanding the Trade-offs
High write quorum (W close to N): Strong durability, but writes fail if many nodes are unavailable. A W=3 in a N=3 cluster means losing any node blocks all writes.
High read quorum (R close to N): Ensures fresh reads, but becomes unavailable if nodes fail. If R=3 and N=3, losing one node means you cannot read.
Balanced quorums (R + W > N): The classic R=2, W=2, N=3 configuration tolerates one node failure for both reads and writes, but requires majority participation.
Low quorums (R=1 or W=1): Maximizes availability—operations succeed as long as any node responds. However, you risk reading stale data (if R=1) or losing writes before replication completes (if W=1).
Real Failure Scenarios
During a network partition, nodes on the minority side of a split may lack quorum. With R=2, W=2, N=3, if two nodes separate from one, the isolated node cannot serve reads or writes—sacrificing availability for consistency.
With sloppy quorums, the system temporarily accepts writes to substitute nodes during failures, prioritizing availability but delaying consistency until failed nodes recover.
When nodes are slow, not crashed, a high quorum requirement increases latency because the system must wait for multiple slow responses.
Key Takeaway: Quorum configurations directly control how your system behaves under failure—higher quorums favor consistency but reduce availability during outages, while lower quorums maximize availability at the cost of potential inconsistency.