In the Grafana U/I you can specify a set of tags for the metric query. We take the first tag and seek to it, then take the remaining tags in the query and filter on them. Specifying the tags in the metric query in order of more specific to less specific will give better performance. I thought I had documented this in the README, but looking for it now I can't find it. Thanks for reminding me :-)
----- Original Message ----- From: "Russ Weeks" <[email protected]> To: [email protected], [email protected] Sent: Wednesday, June 22, 2016 3:42:53 PM Subject: Re: [ANNOUNCE] Timely - Secure Time Series Database Fascinating stuff, thanks! Can I ask, what's the motivation for permuting the tagKey+tagValues between the CF and CQ in the metrics table? It doesn't seem like it would optimize any queries unless you defined locality groups for some or all tagKey+tagValue pairs. In the example in the readme, would you define a locality group for each rack, instance and/or host? Thanks again, -Russ On Wed, Jun 22, 2016 at 11:57 AM <[email protected]> wrote: > > Timely is a time series database application that provides secure access > to time series data. It is designed to be used with Apache Accumulo for > persistence and Grafana for visualization. Timely is located at > https://github.com/NationalSecurityAgency/timely . >
