Metric Targets for Monitoring and Scaling

[akochpaycor]

Based on reading the documentation and other Q&A discussions, it looks like it is possible for YARP to be both CPU or Memory bound.

Do you have any rule-of-thumb thresholds for CPU or Memory that can be used for 1) Scaling Rules on a hosting App Service Plan 2) Warning and Critical level monitoring alerts? (ex. CPU over 90%, Memory over 80% for > 5 minutes)

Are there any other App Service Plan standard metrics that you believe are strongly correlated to YARP health? (ex. Http Queue Length)

[MihaZupan]

I don't believe we have such guidance available, but running YARP should be similar to running any other ASP.NET Core server application.

Are there any other App Service Plan standard metrics that you believe are strongly correlated to YARP health? (ex. Http Queue Length)

If you are exporting .NET metrics for System.Net.Http, counters like http11-requests-queue-duration and http20-requests-queue-duration can be helpful to spot overloaded connections to the backend destinations.