As we grow the next question is around partitioning data and how to cache in front of the db better. Likely based on kind and created_at to trial for partitioning.

I’ve split a fair chunk out, but have jsonb event data in events table too. Size is mostly due to indexes (but also 13MM events) - I should likely time bound them. My workload is less as a relay however.

It will likely come down to what’s your db workload like and what’s the best store for that case - often storing the same data lots of times in different ways (eg. elasticsearch, caching, data warehouse, etc.

Also testing Cassandra for query performance. Need more time.