Examkin

Question

Difficulty: MediumCapacity Planning and Infrastructure Workload Scaling Optimization

An enterprise logistics provider processes real-time vehicle telemetry ingested by a fleet of Compute Engine instances in a Managed Instance Group (MIG) behind an External HTTP(S) Load Balancer. During peak operating hours, incoming message throughput surges by 400%400\%. Load testing demonstrates that while the application is heavily network I/O-bound, the MIG autoscaler is currently configured to scale based on average CPU utilization, causing severe latency spikes before new instances launch. Additionally, project telemetry forecasts indicate that scaling to meet peak load will exceed the project's allocated Compute Engine regional vCPU quota in us−central1us-central1. Which pair of actions should the Cloud Architect recommend to optimize workload scaling and ensure adequate infrastructure capacity?

  1. Submit a regional vCPU quota increase request for us−central1us-central1 prior to peak events, and reconfigure the MIG autoscaler to use a Cloud Monitoring metric based on open network connections or load balancer request count.Answer
  2. B
    Lower the target CPU utilization threshold on the MIG autoscaler to 20%20\%, relying on GCP's automated dynamic quota burst capability to dynamically expand regional vCPU limits during high load.
  3. C
    Migrate the telemetry ingestion service to a GKE cluster with Autopilot mode to bypass regional vCPU quota restrictions while maintaining CPU-based horizontal pod autoscaling.
  4. D
    Provision a 10 Gbps Dedicated Interconnect circuit between the fleet and Google Cloud to bypass load balancer limits, and request a regional vCPU quota increase.

Answer

Submit a regional vCPU quota increase request for us−central1us-central1 prior to peak events, and reconfigure the MIG autoscaler to use a Cloud Monitoring metric based on open network connections or load balancer request count.
For network I/O-bound workloads, CPU utilization remains low even under heavy network load, resulting in delayed autoscaling. Configuring custom metrics via Cloud Monitoring (e.g., HTTP request rates or connection counts) ensures the MIG scales in direct response to network demand. Furthermore, Compute Engine regional vCPU quotas are fixed administrative limits that must be increased in advance via the Google Cloud Console to prevent instance creation errors during scale-out events.

Step-by-Step Solution

1
Evaluate the bottleneck characteristic of the workload
Identified that the application is network I/O-bound rather than CPU-bound, making CPU utilization an ineffective signal for autoscaling.
Autoscaling on CPU utilization for I/O-bound tasks causes delayed scaling and latency degradation.
2
Select an appropriate scaling metric
Configure Cloud Monitoring metrics such as load balancer request count per instance or active TCP connection count.
Custom/I/O metrics trigger autoscaling proactively as incoming network traffic increases.
3
Assess quota constraints
Determine that scaling beyond current limits requires an explicit quota request in advance.
Compute Engine quotas are enforced per region/project and require manual approval before provisioning high instance counts.

Key Concept

Workload-specific autoscaling metric selection and proactive regional quota management
Rate this question

Topics

Capacity Planning and Infrastructure Workload Scaling Optimization