AZ-700 Question 194
Single answerYou manage a client-facing web application hosted on a set of Azure Virtual Machines orchestrated by a virtual machine scale set behind an Azure Load Balancer. The application experiences unpredictable traffic spikes at various times during the day and night. Cost control is essential, but you also want to ensure that the application remains responsive under heavy load. Which scaling approach is most appropriate for this environment?
- A
Configure a fixed number of VMs by setting a manual scale to a moderate count and keep it constant.
- B
Enable autoscale but only schedule it for peak business hours to minimize costs.
- C
Implement dynamic autoscaling rules based on performance metrics (e.g., CPU usage) to scale out and in when traffic spikes or drops.
- D
Disable autoscaling and rely on manual intervention whenever traffic spikes are detected.
Show answer and explanation
Correct answer: C
Explanation
Dynamic autoscaling based on real-time performance metrics allows Azure Virtual Machine Scale Sets to meet varying demand while optimizing costs. According to Microsoft documentation (see Azure Scale Sets documentation), metrics-driven rules ensure that additional instances are added automatically during high demand, and scaled in when demand decreases, preventing both over-provisioning and under-provisioning.
- A. Incorrect.
Option 1 is incorrect. A fixed manual scale might waste resources when demand is low and cause performance issues when demand unexpectedly spikes, because it does not adapt to changes in workload.
- B. Incorrect.
Option 2 is incorrect. Scheduled autoscale is more efficient when traffic patterns are predictable, but in this scenario traffic spikes are unpredictable. Relying solely on a schedule may lead to under-provisioning or over-provisioning.
- C. Correct.
Option 3 is correct. By using metrics-based autoscaling, the scale set will automatically add or remove instances as needed, helping control costs during low demand and ensuring sufficient capacity during spikes.
- D. Incorrect.
Option 4 is incorrect. Relying on manual intervention for scaling will lead to slower response to traffic spikes and increased risk of performance degradation, especially if spikes occur at unpredictable times.