Under 'Detect Categorical Outliers' (sgname is the dimension of the name of logical device groupings). The mstats query gathers write response time for a number of device groupings.
| mstats avg(write_response_metric_name) WHERE index=myindex span=5m by sgname
My question is how can I detect outliers accurately but still know which group it's associated with? If I select the sgname and the metric it returns a very different result than just selecting the metric itself.