Examkin

Question

Difficulty: EasyAuto-scaling and Capacity Planning

A company runs an I/O-bound web service on a Compute Engine Managed Instance Group (MIG). During traffic spikes, request latency increases due to network socket exhaustion, while average CPU utilization on the instances remains below 30%. Which autoscaling configuration should you implement to scale the instances effectively?

  1. Configure the MIG autoscaler using a custom Cloud Monitoring metric that measures request queue depth or active connections.Answer
  2. B
    Configure the MIG autoscaler using a standard target CPU utilization threshold of 80%.
  3. C
    Migrate the web service to Google Kubernetes Engine (GKE) with horizontal pod autoscaling to handle microservice scaling automatically.
  4. D
    Maintain static instance counts on the MIG and request a regional compute quota increase right before traffic spikes occur.

Answer

Configure the MIG autoscaler using a custom Cloud Monitoring metric that measures request queue depth or active connections.
For I/O-bound applications, traditional CPU utilization metrics do not reflect true workload pressure. Exporting custom metrics such as request queue depth or connection counts to Cloud Monitoring enables the Managed Instance Group autoscaler to scale based on the actual resource bottleneck.

Step-by-Step Solution

1
Identify the workload bottleneck.
The application is I/O-bound (network connections/queue depth), so CPU usage remains low during saturation.
Standard CPU metrics will not trigger scale-out events when CPU utilization stays below threshold.
2
Select an appropriate autoscaling signal.
Export application queue depth or active connection metrics to Cloud Monitoring and target those metrics in the MIG autoscaling policy.
Custom metrics accurately signal load for I/O-bound services and trigger timely scale-out events.

Key Concept

Custom metric autoscaling for I/O-bound workloads
Rate this question

Topics

Auto-scaling and Capacity Planning

More questions from this topic