What is split-brain, and why is it dangerous in a replicated database?
Split-brain happens when a network partition causes two nodes to each believe they are the sole leader, both accepting writes independently. When the partition heals, the two histories have diverged and must be reconciled — and any reconciliation strategy risks silently dropping or overwriting real writes. It's the sharpest failure mode of leader-based replication under network partitions.
Answered in
Database Replication Explained: Single-Leader, Multi-Leader, and LeaderlessWhat replication lag actually breaks, and how single-leader, multi-leader, and leaderless models trade off consistency for availability.
Read the full analysisOther questions this article answers
More system design & architecture questions
- What does CodeDeploy RESTART actually deploy?
- Does CodeDeploy RESTART work with ECS or Lambda?
- Does RESTART remove an application from load-balancer traffic while it restarts?
- Does Elastic Beanstalk's Active Directory join require a domain password?
- Can an Elastic Beanstalk environment be healthy if the domain join fails?
- Will enabling Active Directory replace running instances?
- What is different about AWS Lambda Web Functions?
- Does Lambda Web Functions automatically cost less than ordinary Lambda?
Every answer on Crashtech is written by the editor of the article it comes from — never auto-summarised. Browse all answers or the System Design & Architecture beat.