Bartlomiej Plotka

> @bwplotka We've discussed this question in the histograms sync meeting. It was pointed out that we already keep the bucket definitions separate from the values, since the bucket definition...

Related https://github.com/prometheus/prometheus/issues/15200

To progress here: A) We could start by adding a single different record type for RefHistogram to replace float and normal histogram, but also supports custom buckets -- all with...

Agree with the ideas in https://github.com/prometheus/prometheus/issues/14730#issuecomment-2435094351 As per later comment on optimizations based on what changed I think it's def powerful but it feels we have to first DRY the...

I think this can be close for now though as done. Thanks!

Added PRW retrieval of CT from metadata in https://github.com/prometheus/prometheus/pull/15715, benchmarking it now. Outlining clear pros and cons here: https://docs.google.com/document/d/1xbegVBLlDorm0WPuoBR5dNuKJL-I6gp28s-An2ugAME/edit?tab=t.0

Relevant discussion about metadata store: https://cloud-native.slack.com/archives/C01AUBA4PFE/p1724280552347379

One perspective is to wait for this new metadata storage, but actually using current storage might be sufficient for majority of cases, so perhaps a short POC CT might be...

Changed the title to not suggest a solution - we can use either metadata or modify the sample record even.

Some pre-optimization benchmark, built on top of @bboreham work in https://github.com/prometheus/prometheus/pull/15720 ``` benchstat -col /fmt scrape-loop-v1.txt goos: darwin goarch: arm64 pkg: github.com/prometheus/prometheus/scrape cpu: Apple M1 Pro │ PromText │ OMText...