I need to do fast querying over collections with frequent changes. Fortunately, the size of these collections won't get too big. Also, I can afford to lose some data, and the architecture does not need to be fault tolerant. The biggest requirement is to increase transaction throughput to the maximum. I have 2 servers, one with a goodProcessor server and the other one smallProcessor server.
I am exploring setting:
- the master db as in-memory storage on goodProcessor server
- the replica of the master db as WiredTiger storage on smallProcessor server
Is this replication setting possible?