Examkin

Soru

Zorluk: ZorSelecting and Designing Compute and Application Platforms

An autonomous vehicle simulation enterprise is designing its cloud architecture on Google Cloud for two distinct workload components:

1. Workload A: A stateless HTTP REST API endpoint that receives vehicle telemetry webhooks with highly unpredictable, bursty traffic patterns. The requirement specifies automatic scaling down to zero instances during idle periods and zero operational overhead for cluster plane or OS management.
2. Workload B: A specialized simulation engine that requires custom kernel parameter tuning (`sysctl`), raw non-HTTP TCP socket listening, and direct access to local NVMe SSD storage.

Which TWO platform architectural decisions should the Cloud Architect recommend? (Select TWO.)

  1. Deploy Workload A on Cloud Run to achieve stateless HTTP container execution with scale-to-zero capabilities and minimal operational management.Cevap
  2. Deploy Workload B on Compute Engine virtual machine instances configured with attached local SSDs to support custom OS kernel modifications and raw TCP socket networking.Cevap
  3. C
    Deploy Workload A on a standard Google Kubernetes Engine (GKE) cluster with dedicated node pools to handle the stateless HTTP webhook requests.
  4. D
    Deploy Workload B on App Engine Standard environment using custom runtimes to configure `sysctl` kernel parameters and attach local NVMe SSDs.
  5. E
    Deploy Workload A on Compute Engine Managed Instance Groups (MIGs) backed by 3-year Committed Use Discounts to minimize cost during zero-traffic idle periods.

Cevap

The optimal solution requires deploying Workload A on Cloud Run for stateless serverless HTTP execution with scale-to-zero capabilities, and deploying Workload B on Compute Engine instances with local NVMe SSDs to accommodate custom kernel configuration and raw TCP networking requirements.
Cloud Run is the ideal fully managed serverless platform for stateless HTTP webhooks because it automatically scales down to zero when idle and eliminates OS and cluster maintenance overhead. Compute Engine is required for workloads that demand OS-level access, such as custom kernel parameter tuning (`sysctl`), non-HTTP TCP sockets, and direct local NVMe SSD attachment.

Adım Adım Çözüm

1
Analyze the requirements for Workload A
Identified stateless HTTP webhooks, bursty traffic, scale-to-zero requirement, and zero operational infrastructure management.
Cloud Run is purpose-built for fully managed stateless container workloads serving HTTP/HTTPS requests with native scale-to-zero capabilities.
2
Analyze the requirements for Workload B
Identified requirement for custom Linux kernel parameters (`sysctl`), non-HTTP raw TCP socket networking, and local NVMe SSD attachment.
Compute Engine provides full administrative access to the virtual machine operating system, kernel parameters, custom protocol networking, and hardware attachment.
3
Evaluate and eliminate unsuitable platform choices
GKE introduces unnecessary cluster overhead for simple webhooks (Workload A), App Engine Standard prevents OS kernel tuning (Workload B), and Compute Engine MIGs incur idle costs for bursty workloads.
Aligning compute platforms to specific operational boundaries optimizes both cost and administrative efficiency.

Anahtar Kavram

Selecting GCP Compute Platforms based on operational overhead, container statefulness, OS/kernel customization needs, and autoscaling profiles.
Bu soruyu puanla

Konular

Selecting and Designing Compute and Application Platforms