The one question
How do you build a storage system that never refuses a write, even while servers crash and networks split?
Eventual consistency means every copy of the data ends up the same once updates stop, though copies may disagree for a while.
The five Cs
- Category
- An industrial system paper: the design of a production database and lessons from running it.
- Context
- Builds on consistent hashing from peer-to-peer systems and on vector clocks from distributed systems theory.
- Correctness
- Assumes servers fail often and networks split, but nobody is malicious. Availability comes first; replicas may briefly disagree.
- Contributions
- Shows that known techniques, combined carefully, give a store that always accepts writes and lets each application tune its trade-offs.
- Clarity
- Clear and practical. Dense in section 4, where each subsection adds one technique.