Ask HN: Cheapest way to run a time-series database in cloud?
I need to run a database (preferably Postgres-based, currently using TimescaleDB) to store about 20M rows of price data per day, with the option to discard or offload data after 7d to cold storage.
It's just a hobby project that doesn't make me any money, so I want to run this as cheaply as possible. I was experimenting with TimescaleDB cloud but the pricing there starts at $40 a month not including storage.
I tried running it myself on Fly.io but the DB imploded within hours of setting it up and was extremely difficult to recover, so I had to give up on that.
Highly recommend clickhouse. You can set the partition key to be based on date and delete old data in chunks. Makes life really easy. Also found it much more performant than timescaledb out of the box. Not sure about hosted options but can imagine that translates to better performance for the dollar.
Is the cloud a must? What kind of performance are you after? You could repurpose an old desktop computer / server, slap a 4TB HDD on it and have it run at home or in your office.
The longer you use it, the cheaper it gets.
I know you said preferably Postgres based, but influxdb has strictly usage based pricing so that might be worth a look. Also you're an open source project and have tight budgetary constraints sometimes you can always ask for a discount, the worst they can say is no.
https://www.influxdata.com/influxdb-cloud-pricing/
Probably Cloudflare's Workers analytics engine (basically serverless clickhouse)
Clickhouse on a cheap hetzner vm or server?
Prometheus on a decent digital ocean box? A $20/mo vps will handle that load easily, though you may need to buy extra storage depending on data shape.
If you're considering a VPS I'd also check out Hetzner. They only have Germany and Finland locations, but if latency isn't an issue (or if you're in Europe) you can get really good prices there.
checkout ClickHouse cloud maybe they have cheaper options for your usage