[cedricd]

Yeah, I think using Snowflake or BigQuery or something is ultimately the better move. But sometimes folks use what they know (what they're comfortable managing, tuning, deploying, whatever).

In my own testing PG performed very similarly to a 'real' warehouse. It's hard to measure because I didn't have the same datasets across several warehouses. Maybe in the future I'll try running something against a few to see.

[fiddlerwoaroof]

I really wanted to migrate an analytics project I was working on from Elasticsearch to Postgres: however, when we sat down and ran production-scale proofs of concepts for the change, ClickHouse handily outclassed all the Postgres-based solutions we tried. (A Real DBA might have been able to solve this for us: I did some tuning, but I’m not an expert). ClickHouse, however, worked near-optimally out of the box.

[willvarfar]

Snowflake and big query are cloud solutions. Some companies have a need for self hosted databases etc.

And often having a homogenous database stack is a plus. If your production systems are all MySQL, then trying to get away with using MySQL for analytics too is a smart move etc.

I’ve seen so many tiddly data warehouses. Most companies don’t need web scale, and they overbuild and over complicate when they could be running on a simpler homogenous stack etc.