Anyone have experience running their own time series database for exposing analytics back to customers (i.e. not purely for internal use). 

Thinking about giving frame developers much more detailed analytics on their frames, and exploring one of a few options.

Anyone have experience with time series dbs? 

Working on some infra for frame devs

Technowatermelon. Elder Millenial. Building Farcaster. 

nf.td/varun

🙋 we built this https://www.pinata.cloud/blog/how-to-use-farcaster-frame-analytics

Building @warpcast and @farcaster. Previously Coinbase, Brigade, Causes.

I’ve used Timescale DB for time series data for analyzing robotics sensor data @sds via grafana etc for ROS / MCAP data files (no reason it couldn’t be Parquet either, just didn’t build a parquet driver) - looked into QuestDB for a bit as well

fiids.xyz | deescuss.com | bondes.xyz | feedbac.xyz

bringing onchain communities together • events.xyz (@event) 🍊

👋 most important thing I’ve learned is that the selection of a timeseries DB highly depends on the use case. benchmarked most of them approximately 3 years ago and ended up selecting clickhouse. happy to chat more about it, differences between them, how to operate at scale, pros & cons, etc @v @sds

Timescale is really really well done and probably one of the best fits for all but the most demanding applications. i managed a multi-trillion “row” cluster for 5+ years

I do, are you looking for some comparison of options?

Yup.  We're working with Timescale DB rn for something that will be user-facing

I use duckDB in python for storing hourly power market time series data. It's definitely been flexible enough for what I have needed