Hi,

1. Discord uses Cassandra as an actual store for their messages.

2. For the syncing part, I think there are various mechanisms you can do this. When you are putting a message in the main data source, publish that message to a distributed queue also, which will then later be consumed by Elastic search and finally be indexed.

Some databases like dynamodb also offer a mechanism of streams in which, for all the records which are written, a stream is generated and that stream can be consumed to be ingested into Elastic Search.