I have a log-based metric with a high cardinality label. When I try to graph it over a time period with ~5-10k unique values, Monitoring struggles to include them all individually:
Is there any way to break out just the top N values and collapse the rest? Preprocessing function, group by parameter, anything else? Happy to use MQL if necessary.
(I know this is a nontrivial problem in the general case, streaming algorithms etc, but I’m fine with any approximation that works and doesn’t hang my browser for 10s of seconds at a time. 😁)