what is the best approach to display on dashboard just those ~15 method calls out of 40 000/day that are slow = crossed the configured threshold?
If you want to control only the average response time of the methods you can work with one MG that includes the 15 metrics and you can visualize the related alarm through graph obiect in dashboard, showing Minimun and Maximum values.
To avoid losing evidence of the worst calls, due to the huge number of method calls and the average operator, you could configure one calculator for each method that extracts the max value for each interval. Then you could configure a MG that includes the custom metric agent virtual metrics generated by the calculators and set and visualize the related alert in dashboard. This is and expensive solution, but gives you more information about the methods behavior.
Was Marco's response helpful?
Do others have anything else to add?
Retrieving data ...