We developing a service and the target datastore for my service is Cassandra. I want to use spring data flow for ingesting feed. As data flow server requires RDBMS to keep track of workflow, i am forced to choose an RDBMS (eg:postgres) and need to provision it for SDFS. My question is that should DB(used by SDFS) be highly available & scalable to keep my service up/running on target runtime(eg: kubernetes)?
Dataflow usage of DB is very light. It does not mean you should not provide a high available DB in production though. As far as it concerns to scaling concerns it's very light on accessing the DB so you don't need to worry about it being a burden on it. It only uses db to store metadata not during the ingestion of data over the streams