Content
@
0 reply
0 recast
0 reaction
Shane da Silva
@sds
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.
9 replies
1 recast
20 reactions
Justin Hunter
@polluterofminds
Talk to @mattober. He set up our time series database for Piñata’s frame analytics.
1 reply
0 recast
2 reactions
Shane da Silva
@sds
@mattober I'm just looking for a high level synopsis—what's the operational experience, developer experience, and if you have any anecdotal data, performance experience based on the stack you chose. No judgements, just curiosity. Thanks!
1 reply
0 recast
0 reaction
Matt Ober
@mattober
We ended up choosing a clickhouse based system for our time series needs. Both data ingestion speed and query performance are freakishly good. We also liked the “SQL-like” query language.
2 replies
0 recast
2 reactions
Matt Ober
@mattober
Like a lot of modern DBs, operational experience needed kind of depends on whether you buy vs build out your stack. You can run clickhouse yourself, but there’s also a lot of “out of the box” cloud-hosted tooling saas options that are fairly easy to pick up
1 reply
0 recast
2 reactions
Matt Ober
@mattober
The gotchas depend a bit on your specific use case, but the main thing I would mention is that design requires a bit more design thinking up front as tables are essentially append only and you can’t modify things like indexes or constraints after table creation. Happy to chat further if you have deeper questions
0 reply
0 recast
2 reactions
