- MongoDB replicaSet - ES Sharded
- Sharded MongoDB - ES Sharded
Infrastructure level sharding
App Level Sharding
As much as possible, we want each of the "sharding" step to provide a smooth transition.
In the context of a real-life deployment, it may not be realistic to expect that to go from x Millions to y Billions, we need to fully export/reimport data.
This means that we need to identify:
- migrate MongoDB from simple ReplicaSet to Sharded Cluster
- impact of adding shards to an ES cluster
- Shard MongoDB only on UID?
- Gain from multi-collections in the same MongoDB cluster?
- for ES when do we add shards?
- Should we simply maximize the number of shards from the beginning