Soru

Zorluk: OrtaCapacity Planning and Infrastructure Workload Scaling Optimization

A SaaS provider offering real-time document collaboration tools is planning a major enterprise product release expected to cause a 5x surge in API traffic overnight. The application tier runs on Compute Engine Managed Instance Groups (MIGs) handling I/O-intensive requests, backed by a standard Cloud SQL database. During load testing, engineers noticed that CPU load remains low even when network request queues back up, and compute capacity risks reaching default GCP project caps during peak hours. Which TWO actions should the Cloud Architect execute during capacity planning to ensure seamless infrastructure workload scaling? (Select TWO.)

  1. Submit preemptive requests for regional Compute Engine vCPU quota increases across all target regions prior to the launch.Cevap
  2. Reconfigure the Managed Instance Group autoscaling policies to scale based on HTTP request capacity or custom I/O queue metrics rather than CPU utilization.Cevap
  3. C
    Migrate the stateless web application tier from Compute Engine Managed Instance Groups to a multi-cluster Google Kubernetes Engine (GKE) environment to manage scaling overhead.
  4. D
    Replace the single-region Cloud SQL database with Cloud Spanner to scale standard localized user session queries.

Cevap

The Cloud Architect should request regional vCPU quota increases in advance and reconfigure the MIG autoscaling policy to use HTTP request rate or custom I/O queue metrics instead of CPU utilization.
Preemptive regional quota increases prevent automated scaling actions from failing when peak demand hits GCP quota limits. Simultaneously, aligning autoscaling metrics with request rates or custom I/O queue depth ensures that I/O-bound application tiers scale out promptly even when CPU utilization remains low.

Adım Adım Çözüm

1
Analyze workload scaling bottlenecks and quota limitations.
Identified that I/O-intensive workloads do not saturate CPU, making CPU-based autoscaling ineffective, and determined that default GCP regional quotas could cap maximum instance scaling during peak demand.
Ensures scaling triggers appropriately and prevents capacity failure due to quota caps.
2
Select appropriate capacity planning metrics and quota management actions.
Configured custom metrics for MIG autoscaling and submitted advance regional vCPU quota requests.
Allows infrastructure to scale seamlessly in response to actual request load while guaranteeing cloud resource headroom.

Anahtar Kavram

Capacity Planning and Infrastructure Workload Scaling Optimization
Bu soruyu puanla