Question

Difficulty: EasyConfigure Virtual Machine Scale Sets (VMSS)

A web application runs on an Azure Virtual Machine Scale Set named `ecom-vmss`. You configure autoscale rules where scale-out occurs when CPU usage exceeds 75%75\% and scale-in occurs when CPU usage falls below 70%70\%. During periods of normal operation, the scale set repeatedly adds and removes instances in rapid succession. You need to resolve this issue. What should you do?

  1. Decrease the scale-in metric threshold to a lower value, such as 40%40\%.Answer
  2. B
    Change the upgrade policy of the scale set from Automatic to Manual.
  3. C
    Change the orchestration mode of the scale set to Flexible.
  4. D
    Configure the scale-in policy to DeleteNewest.

Answer

Decrease the scale-in metric threshold to a lower value, such as 40%40\%.
Decreasing the scale-in metric threshold (for example, to 40%40\%) creates a wider gap between the scale-out threshold (75%75\%) and the scale-in threshold. This buffer prevents the scale set from immediately deleting newly created instances when the average CPU load decreases post-scaling, thereby stopping the thrashing cycle.

Step-by-Step Solution

1
Analyze the symptoms of the scale set repeatedly adding and removing instances (thrashing).
Identify that the scale-out threshold (75%75\%) and scale-in threshold (70%70\%) are too close to each other.
When a new VM instance is added to the scale set, the average CPU load drops. Because the scale-in threshold is set to 70%70\%, this drop immediately triggers the scale-in rule, leading to a loop of constant scaling.
2
Determine the resolution for autoscale thrashing.
Increase the gap between scale-out and scale-in thresholds by decreasing the scale-in metric threshold.
Setting the scale-in threshold to a lower value (like 40%40\%) provides a sufficient buffer, ensuring that instances are only terminated when workload demand has significantly decreased.

Key Concept

Autoscale thresholds and thrashing prevention
Estimated Time:1m 0s
Rate this question