It would be great to support both legacy histograms and new native histograms. This will provide a powerful data source for better analysis. Native histograms will be especially nice as these ping workloads can vary and defining buckets by hand is hard to size correctly.
Native histograms should be feature flagged and default off due to their experimental nature.
For regular histograms, consider metric creep as 10 buckets per target could get expensive fast.. maybe feature flag that too
It would be great to support both legacy histograms and new native histograms. This will provide a powerful data source for better analysis. Native histograms will be especially nice as these ping workloads can vary and defining buckets by hand is hard to size correctly.
Native histograms should be feature flagged and default
offdue to their experimental nature.For regular histograms, consider metric creep as 10 buckets per target could get expensive fast.. maybe feature flag that too