Definition
Anti-entropy, gossip, and read repair are convergence mechanisms that continuously reconcile replica divergence in eventually consistent systems.
Key Concepts
- Gossip disseminates membership and update knowledge.
- Anti-entropy performs background state reconciliation.
- Read repair fixes stale replicas during read paths.
- Convergence speed depends on topology and load.
How It Works
Nodes exchange digests or Merkle-tree summaries, identify divergence, and stream missing updates; reads can opportunistically repair stale replicas.
Comparison Table
| Mechanism | Path | Primary purpose |
|---|---|---|
| Gossip | Background | State dissemination |
| Anti-entropy | Background | Replica convergence |
| Read repair | Read-time | Opportunistic correction |
Production Implications
Dynamo-inspired systems and Apache Cassandra rely heavily on these mechanisms. Poor tuning causes long-tail inconsistency windows, compaction pressure, and elevated repair traffic during incidents.
When to Use / Not Use
Use It When
- Large replicated clusters with asynchronous replication.
- Systems prioritizing availability during partitions.
Avoid It When
- Tiny single-region clusters where synchronous replication is cheap.
- Strictly linearizable workflows.
Key Takeaways
- Start with user-visible correctness requirements, then choose the weakest model that still preserves them.
- Instrument staleness and conflict rates; treat them as first-class production metrics.
- Design reconciliation and repair workflows before incidents force ad-hoc fixes.