Replication and fault tolerance MongoDB
Managed Service for MongoDB supports replication by default: if a cluster has multiple active hosts, the first replica that handles write requests is automatically selected out of them.
Due to limited resources, the b1.nano, b1.micro, b2.nano, and b2.micro class hosts are not replicated.
For more information about how replication works in MongoDB, read the DBMS documentation.
For cluster hosts to be able to automatically select the primary replica if needed, the vast majority of hosts need to be operable. As a result, it's more cost-efficient to deploy clusters with an odd number of hosts when working with Managed Service for MongoDB. For example, a cluster with 3 hosts can't lose more than 1 host and keep working. At the same time, a cluster with 4 hosts also can't lose more than 1: if a second host is lost, the remaining hosts won't be enough to select a new primary replica.
A cluster with 2 hosts doesn't ensure full fault tolerance for the same reason: a single remaining host can't select itself to be the primary replica. In this situation, the cluster can only process read operations.