You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jan 29, 2025. It is now read-only.
It would be useful to have namespaces for metrics, in particular for multi-tenancy. This would allow us to have usage (in bytes and points/sec) per namespace.
Since most users [citation needed] already have some sort of namespacing based on the first 1~3 elements of the metric paths (e.g. app.component, team.app.component, etc.),
maybe we could start by using a simple prefix tree, possibly with configurable depth?
It makes it low-cost to compute stats per node and merge them, as long as the only goal remains computing stats.
With have unfertonately both team.subsystem and bla.team.subsystem. It will make it hard to group by 'team' if we just look at the first three elements. (Unless we group usage for all the combinations of the first three elements, but that may be a lot)
It would be useful to have namespaces for metrics, in particular for multi-tenancy. This would allow us to have usage (in bytes and points/sec) per namespace.
Implentation wise, for carbon we could probably reuse some of that https://github.com/criteo-forks/carbon/blob/master/lib/carbon/storage.py
For Graphite we could add a way to filter out things that are not in the current namespace, but that isn't super important for now.
The text was updated successfully, but these errors were encountered: