Examkin

Question

Difficulty: Very hardSelecting and Designing Compute and Application Platforms

A digital media enterprise is redesigning its cloud media processing architecture on Google Cloud to handle two distinct application workloads:

1. Ingestion Webhook Service: A stateless microservice handling high-burst HTTP/2 traffic from global content providers. It requires rapid scaling to zero, zero infrastructure management overhead, and low per-request processing latency.
2. Video Transcoding Engine: A legacy batch rendering application that requires custom Linux kernel parameters (`sysctl`), proprietary C++ libraries, specialized NVIDIA GPU driver installations, and direct POSIX filesystem mounts for block storage.

Which TWO platform architectural choices should the Lead Cloud Architect recommend to satisfy the technical requirements while optimizing for cost efficiency and operational overhead?

  1. Deploy the stateless Ingestion Webhook Service on Cloud Run to leverage serverless automatic scaling and eliminate cluster control plane management.Answer
  2. Deploy the Video Transcoding Engine on Compute Engine Managed Instance Groups (MIGs) using custom VM disk images configured with the required kernel modifications and GPU drivers.Answer
  3. C
    Deploy the stateless Ingestion Webhook Service on a dedicated Google Kubernetes Engine (GKE) Enterprise cluster to ensure maximum infrastructure control and custom ingress routing.
  4. D
    Deploy the Video Transcoding Engine on Cloud Run jobs mounted to Cloud Storage buckets using FUSE to execute the custom kernel configurations and GPU driver modules.
  5. E
    Configure the Video Transcoding Engine compute instances to autoscale exclusively based on CPU utilization and purchase 3-year Committed Use Discounts (CUDs) for peak capacity.

Answer

The optimal architectural strategy is to deploy the stateless Ingestion Webhook Service on Cloud Run and deploy the specialized Video Transcoding Engine on Compute Engine Managed Instance Groups (MIGs) using custom VM images.
Cloud Run is the optimal target for stateless HTTP microservices that require rapid scaling to zero and minimal operational maintenance. Compute Engine MIGs with custom images provide the required OS-level flexibility for custom kernel tuning (`sysctl`), specialized GPU driver installations, and direct POSIX block storage attachments.

Step-by-Step Solution

1
Evaluate the Ingestion Webhook Service requirements.
The service is stateless, uses HTTP/2, experiences bursty traffic, and mandates zero operational management.
Cloud Run provides fully managed serverless container execution that scales to zero and eliminates infrastructure node maintenance.
2
Evaluate the Video Transcoding Engine requirements.
The application requires custom sysctl kernel settings, specialized GPU drivers, custom binaries, and POSIX filesystem access.
Fully managed serverless platforms restrict host kernel modifications and custom driver installations, making Compute Engine MIGs with custom images the appropriate platform choice.
3
Identify misallocations in distractor options.
GKE introduces unnecessary cluster lifecycle overhead for simple HTTP microservices, while Cloud Run cannot support host-level kernel driver modifications.
Selecting the right GCP compute service requires balancing operational control against management burden based on workload statefulness and OS-level dependencies.

Key Concept

Selecting GCP Compute Platforms Based on Kernel Customization, Statefulness, and Operational Overhead
Rate this question

Topics

Selecting and Designing Compute and Application Platforms