Examkin

Question

Difficulty: HardSelecting and Designing Compute and Application Platforms

A global media enterprise is redesigning its video processing pipeline on Google Cloud. The architecture requires deploying two distinct workloads:

1. A legacy video transcoding service that depends on custom OS kernel parameter tuning (`sysctl` network buffer settings) and requires POSIX-compliant high-performance local disk access.
2. A lightweight REST API service that receives unpredictable, bursty HTTP traffic, requires automatic scaling to zero instances to minimize costs during idle hours, and must require minimal operational management overhead.

Which TWO compute platform architectural decisions should the Cloud Architect recommend to satisfy these requirements? (Select TWO.)

  1. Provision Compute Engine virtual machines using Managed Instance Groups for the legacy video transcoding service.Answer
  2. Deploy the REST API service on Cloud Run.Answer
  3. C
    Deploy the REST API service on a standard Google Kubernetes Engine (GKE) cluster.
  4. D
    Deploy the legacy video transcoding service on Cloud Functions (2nd gen).
  5. E
    Deploy both workloads onto App Engine Flexible Environment to standardize the deployment environment.

Answer

The optimal architecture combines Compute Engine virtual machines (in Managed Instance Groups) for the OS-dependent transcoding workload and Cloud Run for the stateless, bursty REST API service.
The solution requires matching specific compute characteristics to workload demands. Provisioning Compute Engine virtual machines is necessary for the transcoding engine because it grants root access to modify Linux kernel `sysctl` network buffer parameters and attach high-throughput local storage. Deploying the REST API on Cloud Run is optimal because it natively handles HTTP request traffic, automatically scales to zero during inactivity, and eliminates server management overhead.

Step-by-Step Solution

1
Analyze the technical requirements of the first workload (legacy transcoding engine).
Identified rigid OS dependencies: need for custom kernel `sysctl` tuning and local POSIX storage performance.
Serverless and managed application platforms (Cloud Run, Cloud Functions) restrict OS-level kernel tuning and raw block/local POSIX storage configurations, necessitating Infrastructure as a Service (Compute Engine).
2
Analyze the technical requirements of the second workload (REST API gateway).
Identified serverless fit: stateless HTTP handling, unpredictable spikes, scale-to-zero requirement, and zero operational overhead goal.
Cloud Run natively satisfies HTTP request handling with automatic scaling down to zero instances and zero cluster lifecycle management.
3
Evaluate and eliminate candidate compute misallocations.
Rejected GKE for the lightweight API due to management overhead, and rejected serverless options for the transcoding engine due to OS constraints.
Architectural selection must balance feature support, operational simplicity, and cost efficiency based on Google Cloud best practices.

Key Concept

Selecting and Designing Compute and Application Platforms
Rate this question

Topics

Selecting and Designing Compute and Application Platforms