PCA Question 66

Community Discussion

Written by other learners. Comments can disagree with the marked answer — check the official documentation when in doubt.

• 1

  1. Inefficient Resource Utilization (f1-micro): The f1-micro machine type is a shared-core VM, meaning its CPU is a fraction of a physical core and offers burstable performance. The application needs a minimum of 0.1 CPU cores. While an f1-micro can provide this, it is not designed for sustained production workloads and can be unreliable under load. This machine type is best suited for low-traffic websites or small-scale testing, not a production application that needs to maximize utilization. A single f1-micro is a poor fit for hosting a production application that requires stability.

• Chose C2

  The core of the question is not just to get the application running, but to "monitor and maximize machine utilization" and to "reliably deploy new versions." Maximizing Utilization: An application requiring just 0.1 CPU cores and 128 MB of memory is a perfect fit for a container-orchestration platform. A single n1-standard-1 machine has 1 vCPU and 3.75 GB of memory. In a Managed Instance Group (MIG) with option B, a single application uses only a fraction of those resources, leaving the rest idle. In a GKE cluster (Option C), Kubernetes' scheduler can "bin-pack" multiple pods from the same or different applications onto that same node, significantly increasing resource utilization and cost-efficiency.

• 1

  Reliable Deployments: GKE provides a superior deployment pipeline. Using Kubernetes Deployments offers features like rolling updates, automated health checks, and rollbacks, which are more robust than the instance template updates in a MIG. The container-based workflow is also faster and more flexible for frequent updates. Why B and D Are Flawed Option B (MIG with VM Images): While a common approach, building and deploying new VM images for every minor change is a slower and less agile process compared to building and pushing a new container image. It's a less efficient deployment model. Option D (GKE with Bad Practices): This option is the worst of the bunch. Using an n1-standard-4 machine is massive over-provisioning, completely failing to meet the goal of maximizing utilization. The use of the mutable latest tag and imagePullPolicy: Always is a major anti-pattern in production, leading to unpredictable and irreproducible deployments.