This is the most straightforward method of replication. It can be explained as follows:
- We take a cluster of databases:
Cluster of databases
- Designate one of them as the master, and the remaining databases as replicas:
DB-3 is designated as the master
- All writes are performed to the master:
Three writes are performed on the master
- All reads are performed from the replicas:
>
Reads performed from the replicas
- The master ensures that all the replicas have the latest state which is the state of the master database:
The master updates all the replicas with the latest update
- Failure in master database still allows reads from replicant databases but writes are not possible:
Master failure; no writes, only reads