Question

Difficulty: MediumScale Azure App Service Web Apps

A company hosts a high-traffic e-commerce web application named CartService on Azure App Service. The application runs on a Standard (S1) App Service plan. During flash sales, the application experiences sudden traffic surges, causing incoming requests to queue up before they can be processed by the web workers. You need to configure an Azure Monitor autoscale rule to scale out the App Service plan instances to handle the queued requests. Which metric should you select to trigger the scale-out rule?

  1. A
    Active Session Count
  2. HTTP Queue LengthAnswer
  3. C
    Request Execution Time
  4. D
    CPU Core Count

Answer

HTTP Queue Length
The HTTP Queue Length metric represents the number of requests that have entered the queue but have not yet been assigned to a worker thread. When a web application experiences sudden traffic surges, this queue grows, indicating that the existing instances are fully loaded. Autoscaling based on this metric ensures that additional instances are added to handle the excess request volume.

Step-by-Step Solution

1
Identify the bottleneck and the symptom described in the scenario.
The scenario states that requests are queuing up before being processed by the web workers.
This indicates that the worker threads are fully occupied and cannot process incoming requests fast enough.
2
Evaluate the available built-in metrics for App Service plan autoscaling.
Azure Monitor offers built-in metrics such as CPU Percentage, Memory Percentage, and HTTP Queue Length at the App Service level.
We need to select the metric that directly reflects the queuing symptom.
3
Select the metric that triggers scaling when requests queue up.
HTTP Queue Length directly measures the size of the request queue.
Scaling out when HTTP Queue Length is high ensures that additional web server instances are provisioned to clear the request queue and restore application responsiveness.

Key Concept

Autoscale metrics for Azure App Service plans
Estimated Time:1m 30s
Rate this question