With Sentry 10, as they have added ClickHouse to the stack, why do they need Postgres now. As far as I can understand from this article. They are now storing events and their details in Clickhouse. So, I was wondering what is use case of Postgres now?
ClickHouse isn’t a relational database, Postgres is. So all relational and transactional and mutable data is still in Postgres.
So, raw events are stored in clickhouse and data like types of events, user management etc is stored in postgres? Also, where are aggregated stats stored in, Postgres or ClickHouse?
It depends on what stats and data you’re specifically talking about. These things change and evolve over time.
We try to store less and less aggregated stats since we can compute them on the fly in ClickHouse. That’s the whole point. We still store some stats in Postgres, I believe, and Redis yet.