Spike: Improve RedisHLL metrics aggregation performance
As an outcome of #295204 (closed) we understand we cannot remove the existing slow metrics. So instead we need to find the way to speed up aggregation logic.
Why is it working slow
When we aggregate metrics we need to find union of HLL between set of keys. When amount of keys becoming significant, Redis need more than 0.01s which is considered as a slow operation. More details how we aggregate keys in this comment
What can we do
We shall consider migration daily
to be weekly
Edited by Alina Mihaila