DEV Community

Choosing TSDBs - InfluxDB for Us

Steve Mushero on April 20, 2017

Time-Series Databases are powerful and interesting beasts. We are selecting a new one for our OpsStack Total Operations Platform, for use in all ...
Collapse
 
manigandham profile image
Mani Gandham

Might as well just use something like MemSQL and get a real distributed relational database with columnstore functionality. Works great with time-series or any other data warehouse needs with full SQL support, including joins (and fast in-memory rowstore tables for OLTP).

Collapse
 
stevemushero profile image
Steve Mushero

Yes, InfluxDB certainly needs an HA / distributed solution, but using non-TSDB like this means losing tagging, protocol support (UDP rules!), alerting options, data aggregation, dimensional / time analyses, and probably more like high cardinality, multi-tag indexing on large data sets - especially with some newer TSDBs really diving into stats functions for clustering, anomalies, etc. MemSQL looks interesting, a bit complex, MySQL wire-compat is nice. I'm pretty new to TSDBs, too, so probably missing stuff.

Collapse
 
sdracerunner profile image
Tony

Influx does have a HA solution, but it's closed source and had associated license costs. We use it in our setup with a load balancer to front the HTTP endpoint and it works pretty well. Search for Influx Enterprise for details, all the docs are public.

Thread Thread
 
stevemushero profile image
Steve Mushero

Yes, they do have a new commercial HA system, though seems quite young and I think was created after another attempt, so still a ways to go, plus of course would prefer to see an open-source edition, even if had limited shards, size, etc. so lots of people could run it for real in real systems.

Collapse
 
riklempens profile image
Rik Lempens

If you are interested in a fully open-source time-series database with clustering support build into it's core you might want to take a look at SiriDB, a time-series data we build from the ground up to be scalable on the fly, fast and robust.

The full project is available on GitHub