Examkin

Question

Difficulty: MediumAuto-scaling and Capacity Planning

An enterprise healthcare platform uses Compute Engine Managed Instance Groups (MIGs) to run I/O-bound microservices that ingest and process HL7 FHIR clinical data streams. During peak admission hours, processing delays increase significantly because the instances fail to scale out, even though individual VM CPU utilization remains below 25%. Furthermore, a planned hospital integration next month is projected to triple the baseline event volume. Which combination of actions should the Cloud Architect take to resolve the autoscaling issue and ensure capacity for the expansion?

  1. Configure the MIG autoscaler using a Cloud Monitoring custom metric representing active connection queue length, and submit a regional vCPU quota increase request prior to the planned hospital integration.Answer
  2. B
    Lower the MIG autoscaler target CPU utilization threshold from 80% to 15% to force earlier instance scale-out during ingestion bursts.
  3. C
    Re-architect the workload by migrating the microservice to a Google Kubernetes Engine (GKE) cluster with Horizontal Pod Autoscaler (HPA) configured on CPU memory consumption.
  4. D
    Configure custom metric autoscaling based on queue depth, but defer requesting regional Compute Engine quota increases until the MIG triggers quota limit alerts during production traffic spikes.

Answer

Configure the MIG autoscaler using a Cloud Monitoring custom metric representing active connection queue length, and submit a regional vCPU quota increase request prior to the planned hospital integration.
For I/O-bound applications, scaling metrics based on CPU utilization fail to capture true workload bottlenecks. Utilizing Cloud Monitoring custom metrics (such as active connection queue length or Pub/Sub depth) enables the MIG autoscaler to respond dynamically to load. Furthermore, capacity planning requires requesting compute quota increases in advance to ensure the cloud infrastructure can scale up without encountering platform quota caps during anticipated traffic increases.

Step-by-Step Solution

1
Identify the primary scaling bottleneck for the workload.
The application is I/O-bound and experiences processing delays despite low CPU utilization (<25%), indicating that standard CPU-based autoscaling is ineffective.
I/O-bound microservices bottleneck on network sockets or queue backlogs rather than CPU capacity.
2
Select the appropriate autoscaling metric strategy.
Implement custom metric autoscaling in Cloud Monitoring based on queue depth or open connection metrics.
Custom metrics accurately reflect true workload pressure for non-CPU bound services.
3
Perform proactive capacity planning for future growth.
Request regional vCPU resource quota increases in advance of the anticipated workload expansion.
Autoscaling groups cannot provision instances beyond the project's regional resource quotas, and requesting quota increases in advance ensures seamless scaling during peak events.

Key Concept

Auto-scaling for I/O-bound workloads requires custom telemetry metrics (such as queue depth), while capacity planning mandates proactive resource quota management prior to workload expansion.
Rate this question

Topics

Auto-scaling and Capacity Planning

More questions from this topic