MongoDB perform best when the working set fits all in memory. In large working sets, say larger than 500GB, sharding can be used to aggregate memory across multiple machines. Sharding is complex and expensive. Moreover, fitting the dynamically changing working set size into memory is an elusive and error prone task. Under-provisioning memory leads to massive throughput drop and overprovisioning leads to wasting expensive memory resources.

Plexistor converged memory and storage software solution can deliver high capacity, near-memory speed storage from 1TB to 12TB of capacity, so you don’t need to shard the data or buy expensive memory modules.

You can install Plexistor SDM community edition solution for free on your on-premise server or use a pre-installed version of Plexistor and MongoDB on Amazon AWS cloud infrastructure.