Examkin

Soru

Zorluk: ZorSelecting and Designing Compute and Application Platforms

A healthcare organization is architecting a high-throughput genomic processing platform on Google Cloud. The core containerized pipeline executes compute-intensive analysis jobs that run continuously for up to 4 hours per execution. The pipeline depends on specialized third-party Linux kernel modules that must be loaded directly into the host operating system kernel, and it requires raw block access to high-performance local NVMe SSDs formatted with custom filesystem mount parameters. The organization wants to minimize operational overhead while fully satisfying all technical requirements. Which compute platform should the Cloud Architect select for this workload?

  1. A
    Cloud Run jobs configured with custom memory limits and an extended 4-hour execution timeout
  2. B
    Google Kubernetes Engine (GKE) Autopilot mode deploying Pods with privileged security contexts
  3. Compute Engine Managed Instance Groups (MIGs) using a custom VM OS image pre-configured with the required kernel modules and local NVMe SSD scriptsCevap
  4. D
    App Engine Flexible Environment with custom Dockerfiles configured to load the required kernel drivers upon container initialization

Cevap

Compute Engine Managed Instance Groups (MIGs) using a custom VM OS image pre-configured with the required kernel modules and local NVMe SSD scripts.
Compute Engine provides full Infrastructure-as-a-Service (IaaS) control over virtual machines. When a workload demands specific Linux kernel modules loaded at the host layer and custom local NVMe SSD storage initialization, custom Compute Engine OS images built into Managed Instance Groups provide the required control alongside operational automation.

Adım Adım Çözüm

1
Analyze host-level system dependencies
Identified requirements for third-party host Linux kernel module loading and direct local NVMe SSD custom formatting.
Managed compute abstractions (Cloud Run, App Engine, GKE Autopilot) restrict access to host OS kernel modifications and raw hardware tuning.
2
Evaluate serverless and fully managed compute services against host customization constraints
Ruled out Cloud Run, App Engine Flexible, and GKE Autopilot.
Fully managed compute options lock down node OS kernels for stability and security, preventing custom kernel module injection.
3
Select the optimal compute platform providing OS kernel control with automation capabilities
Compute Engine Managed Instance Groups with a custom OS image.
Compute Engine provides complete root access to VM instances, allowing custom image builds with pre-loaded kernel modules, custom disk mount scripts, and automated lifecycle management via MIG auto-healing and scaling.

Anahtar Kavram

Compute Platform Selection Constraints and Host OS Customization
Bu soruyu puanla

Konular

Selecting and Designing Compute and Application Platforms