It is a method of replication of relational databases where you have 2 master databases that take reads and writes, and coordinate with each other on writes. When one of the masters goes down, the other ones takes the whole load on itself until it goes back up.
Disadvantages
- Requires load balancer or you need to make changes to app logic to determine where to write
- They are loosely consistent (violate ACID) or have increased write latency due to synchronisation
- You have conflict resolution