Dynamo vs Redis Trackstore

My use-case initially does not require persistence for the short term, but the long term approach does require persistence where we should be able to load conversations from 1 year+ ago.

Eventually, we will need dynamo, but we wanted to know if there are any things that Redis bring to the table that ddb falls behind in.

Basically, I am asking, is there value to having a custom track-store that writes to both ddb and redis, where redis is used as a cache layer and ddb is the source of truth, as well as the fall-back.

Or, go straight to ddb without using Redis at all.