Consider switching away from InfluxDB

[Nexarian]

Problem
The dashboard project is frozen on InfluxDB 1.8, and migration to 2.0 feels like a pipedream... There are many more priorities to tackle. If the community is going to tackle a migration, going to something more stable might be better.

Where to migrate to? That I'm unsure of. This issue is to discuss the options.

Enhancement
Eventually the project will need to leave InfluxDB 1.x, and 2.x is not a slam-dunk replacement as it requires a massive re-write/migration. InfluxDB 3.x is on the horizon, and it is AGAIN a full re-write and requires re-writing queries. The marketing materials for InfluxDB 3.0 seem to know that most users are still on 1.8, as they're benchmarking against that, not 2.x: https://www.influxdata.com/lp/oss-vs-new-engine

Doing a full re-write every few years is simply not a viable answer here.

What to use instead?
Here's a useful website that has some alternatives: https://alternativeto.net/software/influxdb

• TimescaleDB

  • Advantage: Based on PostgresSQL, effectively industry standard.

• VictoriaMetrics

  • Advantage: Seems to have an easy drop-in migration path from Influx.

• Prometheus

  • Advantage: CNCF standard. Plays well with the Kubernetes folks.

I've left disadvantages as an exercise for the reader!

[clukas1967]

The only dog I have in the hunt is that any replacement should be available for Synology Docker.

[jasonacox]

The dashboard project is frozen on InfluxDB 1.8

I don't disagree. 😉 Thanks for kicking off a study. I suggest we still consider InfluxDB 3 Core which does hint at being backward compatible. It is in alpha so not ready. But also, good with exploring options.

Doing a full re-write every few years is simply not a viable answer here.

100%. Getting a solution that works with all OS variants (Linux, Win, macOS, RPi) and all platforms (PC, Mac, RPi, Synology, Unraid, etc.) is also a challenge.

I find that most of my time on this project is spent keeping up with Tesla firmware changes or enhancing pyPowerwall for different use cases. The InfluxDB 1.8 part has been remarkably non-dramatic and zero maintenance. That's key because it has allowed us to change out everything around the data without impacting history. I don't deny that it won't need to be upgraded at some point, but I do give it credit for dependability and performance.

[Nexarian]

If InfluxDB ain't broken, don't fix it. Not a bad take. However, the last team I was on at Amazon was to deprecate a 15 year old database that was like that... It was fine, until it wasn't! The amount of pain to eliminate it was... Extreme.

Anyway, I'll also look into Influx3, but probably wait until it comes out of Beta.

Let's postpone evaluation until then.