We are thinking of sharding our database with replication. Our usecases include reads and writes to parts of shards. We have questions like
- How long the delays would be in replication?
- Will there be any data redundancy if one server goes down and comes back (because of delay in replication)?