Summary
Strong consistency means reads reflect the latest committed write. Eventual consistency allows temporary stale reads but converges over time.
Interview Points
- Strong consistency is easier to reason about but can cost latency and availability.
- Eventual consistency improves scale and availability but requires conflict handling.
- Choose per workflow, not globally.
- Financial transfers usually need stronger guarantees.
- Feeds, analytics, search, and counters often tolerate eventual consistency.
2-3 Minute Interview Script
“Strong consistency means once a write succeeds, later reads see it. Eventual consistency means replicas may lag, but the system converges if no new writes occur.
The tradeoff is user semantics. Strong consistency is simpler and safer for critical workflows like payments, inventory reservation, or permissions. But it may reduce availability or increase latency in distributed systems.
Eventual consistency works well when the product can tolerate temporary staleness, like feeds, search indexes, analytics, notifications, or like counts.
In an interview, I would choose consistency per operation. A system can be strongly consistent for account balance updates and eventually consistent for activity feeds.”
Follow-Ups
- What is read-your-writes consistency?
- How do you handle conflicts?