Examkin

Soru

Zorluk: OrtaSelecting and Designing Compute and Application Platforms

A global logistics organization is designing a vehicle telemetry processing system on Google Cloud. The system consists of two distinct workloads: (1) a stateless gRPC event-ingestion microservice that receives high-throughput data streams with unpredictable load spikes, and (2) a high-performance batch simulation service that requires custom Linux kernel sysctl modifications and custom C++ container runtime drivers. The engineering team must minimize infrastructure management overhead and avoid paying for idle compute during off-peak hours. Which TWO compute architecture options should the Cloud Architect select to satisfy these requirements? (Select TWO.)

  1. Deploy the stateless gRPC event-ingestion microservice to Cloud Run using container instances with HTTP/2 enabled.Cevap
  2. Deploy the batch simulation workload to Google Kubernetes Engine (GKE) Autopilot using custom pod security contexts for kernel parameters.Cevap
  3. C
    Deploy the stateless gRPC event-ingestion microservice to a dedicated GKE Standard cluster with fixed worker nodes.
  4. D
    Host the batch simulation workload on App Engine Standard environment to avoid managing container infrastructure.
  5. E
    Deploy the gRPC event-ingestion microservice on Compute Engine unmanaged instance groups behind an Internal Passthrough Network Load Balancer.

Cevap

The optimal architecture combines Cloud Run for the stateless gRPC microservice and GKE Autopilot for the kernel-customized batch simulation workload.
Selecting Cloud Run for the stateless gRPC microservice provides fully managed, zero-to-n auto-scaling with native gRPC/HTTP2 support, meeting the zero idle-cost requirement. Selecting GKE Autopilot for the batch simulation workload accommodates custom Linux kernel (sysctl) settings via pod security contexts while maintaining a serverless operational model where Google manages node infrastructure.

Adım Adım Çözüm

1
Evaluate the stateless gRPC streaming ingestion workload requirements.
Cloud Run is selected because it natively supports HTTP/2 and gRPC streams, automatically scales down to zero to eliminate idle compute costs, and requires zero cluster maintenance.
Stateless containerized services with fluctuating traffic benefit most from fully managed serverless container platforms.
2
Evaluate the batch simulation workload requirements requiring low-level system settings.
GKE Autopilot is selected because it supports specific Kubernetes pod security context configurations (such as sysctl settings) while offloading node lifecycle management and provisioning to Google Cloud.
Workloads requiring low-level kernel parameters cannot run on sandboxed serverless environments like App Engine Standard but still need minimized operational complexity.

Anahtar Kavram

Compute Platform Selection Matrix (Cloud Run vs. GKE Autopilot vs. App Engine)
Bu soruyu puanla

Konular

Selecting and Designing Compute and Application Platforms