logoalt Hacker News

freilanzer05/15/20252 repliesview on HN

Even for timeseries there is https://github.com/timescale/timescaledb. Haven't used it, just knew it existed.


Replies

abraxas05/15/2025

It's very good. Postgres by itself can handle a very high volume of inserts (I did over 100,000 rows/s on very modest hardware). But timescale makes it easier to deal with that data. It's not strictly necessary but it's very time series friendly (good compression, good indexing and partitioning etc). Nothing a pg expert can't accomplish with a vanilla postgres but very, very handy.

show 1 reply
argulane05/16/2025

I have used TimescaleDB in my last work place. We needed a easy way to store and visualize 500hz sensor data for few 10s of devices. We used it and Grafana to build a internal R&D tool and it worked way better than I imagined. Before I left I think the DB was using ~200GB on a compressed btrfs volume in DigitalOcean droplet and still performed fine for interactive Grafana usage.