Several unreliable machines need one ordered decision#
Messages can be delayed, nodes can crash, and every participant has an incomplete view. Consensus protocols define how nodes agree on decisions or an ordered log under a specified failure model, often supporting leader election, metadata, and configuration.
| Replication | Consensus |
|---|---|
| Keeps multiple copies | Agrees on decisions, order, or state transitions |
| May be asynchronous | Uses a defined protocol and failure model |
Replication may use consensus, but the words are not synonyms. Interview depth is the problem, why delay makes failure ambiguous, and what majority agreement enables - not a proof.
Related practice problems
Use the manifest-backed prompts below to apply this lesson in a complete design.
- In Search of an Understandable Consensus AlgorithmUSENIX / Stanford
- Reliability design patternsMicrosoft Azure Architecture Center
Reach this point to mark the lesson as read.