How does Azure know when to scale out?

Question

Hi all;

I understand how to set my Azure App Service to scale out and how I am then charged, What I don't understand is, how does Azure know when to scale out my system? From everything I've read it "just knows." How does it know?

Is it as simple as it only feeds my app N requests at a time, queueing up any additional, and watches that queue size?

Or is there something I need to add to my app to help it?

thanks - dave

Answer 1

Hi Dave,

Depends how you have scaling configured. For example, with Manual you set the number of instances and the count stays constant, regardless of activity.

With Rules-based (aka Custom autoscale) when it scales out/in is based on the rules you create. For example, you could set it so that if CPU percentage stays over 80% for 5 minutes to add an additional instance. CPU percentage is just one example, you can choose from many different metrics.

With Automatic scaling it scales out/in based on incoming HTTP requests. This is relatively new scaling option that requires that you use Premium SKU.

Automatic scaling in Azure App Service

https://learn.microsoft.com/en-us/azure/app-service/manage-automatic-scaling

-TP