Ensure that metrics retention is long enough #211

Open
opened 2025-05-08 17:01:39 +00:00 by raito · 1 comment
Owner

We should make use of downsampling to be able to aggregate long term metrics and keep some information over the 10 years, 5 years, 1 year with different downsampling options.

cc @k900

We should make use of downsampling to be able to aggregate long term metrics and keep some information over the 10 years, 5 years, 1 year with different downsampling options. cc @k900
Owner

Unfortunately, downsampling is currently not implemented nor supported in Mimir.
Most other metrics storage backends however do 🫠

tl;dr for https://github.com/grafana/mimir/discussions/1834#discussioncomment-2727217 and https://github.com/grafana/mimir/pull/5028:

  • s3 storage is cheap
  • downsampling is very IO intensive
  • complicates promql query engine

Most workarounds for this aren't particularly great.
Maybe we defer this for now and lower its priority from "High" to "Medium" or even "Low" and also mark it as blocked (by upstream)?

Unfortunately, downsampling is currently not implemented nor supported in Mimir. Most other metrics storage backends however do 🫠 tl;dr for https://github.com/grafana/mimir/discussions/1834#discussioncomment-2727217 and https://github.com/grafana/mimir/pull/5028: - s3 storage is cheap - downsampling is very IO intensive - complicates promql query engine Most workarounds for this aren't particularly great. Maybe we defer this for now and lower its priority from "High" to "Medium" or even "Low" and also mark it as blocked (by upstream)?
Sign in to join this conversation.
No milestone
No project
No assignees
2 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference: the-distro/infra#211
No description provided.