Examkin

Soru

Zorluk: Çok zorPlanning and Assessing Cloud Compute Solutions

A digital media platform requires an architecturally optimized Google Cloud compute environment to process user-uploaded video files stored in a Cloud Storage bucket. The workload exhibits the following operational characteristics:

• Video processing tasks are triggered dynamically upon file upload.
• Each processing job runs containerized binaries (FFmpeg with custom plugins) taking between 10 to 45 minutes to finish.
• Memory requirements reach up to 16 GB per task execution.
• Traffic patterns fluctuate dramatically, experiencing extreme peak spikes during live events and near-zero activity overnight.
• The engineering team requires zero node management overhead and mandates paying strictly for active compute processing time without idle infrastructure spend.

Which compute solution should be recommended to satisfy these requirements?

  1. Deploy the containerized processing worker as Cloud Run Jobs, triggered via Eventarc events from Cloud Storage, with 16 GB memory allocated per task instance.Cevap
  2. B
    Deploy the processing code as a 1st Generation Cloud Function triggered by Cloud Storage notifications, setting the memory allocation to 16 GB.
  3. C
    Deploy a Managed Instance Group (MIG) of Compute Engine On-Demand instances running 24/7 with 1-Year Committed Use Discounts (CUDs) to cover anticipated peak capacity.
  4. D
    Deploy a Managed Instance Group (MIG) composed of Spot VMs utilizing CPU utilization autoscaling metrics to handle incoming processing queues.

Cevap

Deploying the containerized processing worker as Cloud Run Jobs, triggered via Eventarc events from Cloud Storage, with 16 GB memory allocated per task instance is the optimal solution.
Cloud Run Jobs is designed specifically for containerized workloads that run to completion. It supports execution times up to 24 hours, memory allocations up to 32 GB (or more in current configurations), event-driven invocation via Eventarc from Cloud Storage, zero infrastructure management, and true pay-per-use billing that scales to zero during idle overnight hours.

Adım Adım Çözüm

1
Analyze workload execution duration and runtime dependencies.
Tasks run containerized binaries for 10–45 minutes with 16 GB RAM requirements.
Traditional event-driven functions (like 1st-Gen Cloud Functions) fail due to a 9-minute execution timeout. Cloud Run Jobs supports long-running containerized tasks up to 24 hours.
2
Evaluate infrastructure operational overhead and scaling constraints.
The requirement demands zero node management overhead and serverless billing.
Compute Engine Managed Instance Groups require OS maintenance, patch management, and complex scaling configuration, whereas Cloud Run Jobs is fully managed.
3
Assess cost trade-offs during idle period fluctuations.
Traffic drops to near zero overnight.
Sustained or Committed Use Discounts on 24/7 VM instances incur continuous costs regardless of utilization, whereas Cloud Run Jobs charges $0 when no jobs are executing.

Anahtar Kavram

Selecting serverless container execution paradigms (Cloud Run Jobs vs Cloud Functions vs Compute Engine) based on execution duration, operational overhead, and cost structure.
Bu soruyu puanla

Konular

Planning and Assessing Cloud Compute Solutions

Bu konudan daha fazla soru