Consistency

Consistency is about what a reader can safely assume after a write. Stronger guarantees usually cost latency or availability.

How It Works

Consistency determines what a reader sees after a writer updates data. Strong consistency means every read returns the latest write — simple but expensive across regions. Eventual consistency means replicas converge over time but reads may be stale. In between: read-your-writes (user sees their own updates immediately), causal consistency (events happen in order), and bounded staleness (reads lag by at most N seconds). In interviews, always name which consistency level you chose and why.

Real-World Example

Amazon DynamoDB defaults to eventual consistency for reads but offers strongly consistent reads at 2x the cost. For a shopping cart, eventual consistency is fine. For inventory count during checkout, strong consistency prevents overselling.