Examkin

Question

Difficulty: Very hardCapacity Planning and Infrastructure Workload Scaling Optimization

An online retail enterprise is preparing for an upcoming annual mega-sale expected to drive a projected 10×10\times surge in peak user traffic over a 48-hour window. Their primary application architecture consists of a stateless HTTP API tier deployed on Compute Engine Managed Instance Groups (MIGs) and a Cloud SQL PostgreSQL backend database. During previous smaller sales, sudden traffic spikes caused temporary latency spikes because MIG reactive autoscaling took several minutes to spin up new virtual machine instances, and the Cloud SQL backend suffered from connection exhaustion. Furthermore, the architecture team needs to ensure infrastructure scaling will not fail due to cloud platform limitations during the event. Which capacity planning and workload scaling optimization strategy should the Cloud Architect recommend?

  1. Audit existing project quotas and request regional Compute Engine CPU and IP quota increases well in advance, enable predictive autoscaling or schedule scheduled autoscaling scaling for the MIGs, and introduce connection pooling alongside database read replicas for Cloud SQL.Answer
  2. B
    Configure standard reactive CPU-utilization autoscaling on the Compute Engine MIGs with maximum instance limits set to scale out dynamically as requests arrive, relying on automated GCP dynamic quota allocation during the event.
  3. C
    Re-architect the stateless web application tier into a multi-region GKE Enterprise cluster using Autopilot mode to replace the Compute Engine MIGs, relying on GKE cluster management to resolve regional compute quota constraints automatically.
  4. D
    Migrate the Cloud SQL PostgreSQL database to a multi-region Cloud Spanner instance to automatically eliminate connection pooling limits and solve compute instance autoscaling bottlenecks on the frontend tier.

Answer

The Cloud Architect should recommend proactive capacity planning: requesting regional resource quota increases well in advance, implementing predictive or scheduled autoscaling on Compute Engine MIGs to eliminate startup latency, and optimizing database scaling with connection pooling and read replicas.
Proactive capacity management requires requesting regional compute and network quota increases in advance to ensure the cloud project can accommodate the 10×10\times traffic surge. Combining predictive or scheduled autoscaling on MIGs ensures VM instances are fully provisioned before the traffic arrives, eliminating boot-up latency spikes. Additionally, adding connection pooling and Cloud SQL read replicas addresses the database connection exhaustion bottleneck without requiring expensive re-architecture.

Step-by-Step Solution

1
Evaluate resource quota constraints ahead of high-traffic events.
Identified that GCP quotas (e.g., regional CPU limits, in-use IP addresses) are hard bounds that require advance approval from Google Cloud support.
Failing to request quota increases in advance can cause compute instance creation to fail during rapid scale-out events.
2
Select the optimal compute autoscaling strategy for predictable traffic surges.
Chose predictive autoscaling or scheduled scaling for Compute Engine MIGs over purely reactive autoscaling.
Predictive autoscaling analyzes historical load patterns and provisions capacity ahead of expected spikes, preventing initialization latency and request drops.
3
Address database tier bottlenecks and connection scaling.
Implemented connection pooling (such as PgBouncer) and Cloud SQL read replicas.
Connection pooling multiplexes frontend requests to minimize backend connection overhead, while read replicas offload read query capacity from the primary instance.

Key Concept

Capacity Planning and Workload Scaling Optimization
Rate this question

Topics

Capacity Planning and Infrastructure Workload Scaling Optimization