Then you have the question of how to query it efficiently. You can’t scan 1 year of data every time, so now you need some type of external process that can eit

Then you have the question of how to query it efficiently. You can’t scan 1 year of data every time, so now you need some type of external process that can either pre-compute those metrics (not flexible) or a compaction type process that makes it manageable to scan all that data. Which is basically the trade off between something like stream processing or compaction in a database.