Examkin

Soru

Zorluk: ZorPlanning and Assessing Cloud Compute Solutions

A financial analytics company is planning to deploy a high-performance quantitative risk simulation service on Google Cloud. The application runs as a custom containerized background process requiring custom Linux sysctl kernel parameters to optimize network socket buffers. The workload operates continuously 24/7, requires persistent block storage for local scratch caching, and cannot tolerate sudden instance preemption or termination. The cloud engineering team wants to select the most suitable compute platform while satisfying all operational requirements. Which compute strategy should the team choose?

  1. Deploy the workload on Google Kubernetes Engine (GKE) Standard node pools composed of standard Compute Engine virtual machines.Cevap
  2. B
    Deploy the workload on Compute Engine using Spot Virtual Machines attached to persistent disks to minimize compute costs.
  3. C
    Deploy the containerized workload on GKE Autopilot clusters to eliminate cluster node management overhead.
  4. D
    Deploy the workload as a container image on Cloud Run with persistent disk volume mounts enabled.

Cevap

Deploy the workload on Google Kubernetes Engine (GKE) Standard node pools composed of standard Compute Engine virtual machines.
Google Kubernetes Engine (GKE) Standard node pools using standard Compute Engine VMs offer the necessary flexibility to customize host OS settings—such as modifying Linux sysctl kernel parameters—while providing reliable, continuous 24/7 compute capacity without risk of abrupt preemption.

Adım Adım Çözüm

1
Analyze workload requirements and non-negotiable technical constraints.
Identified constraints: continuous 24/7 execution (no abrupt preemption allowed), custom Linux sysctl kernel parameter tuning requirement, containerized runtime, and persistent block storage.
Defining strict technical constraints prevents choosing serverless or preemptible options that violate kernel or availability mandates.
2
Evaluate serverless and fully managed compute platforms (Cloud Run and GKE Autopilot).
Cloud Run and GKE Autopilot are eliminated because fully managed serverless environments restrict low-level host OS kernel configuration (sysctl tuning).
GKE Autopilot enforces strict node security baselines preventing custom sysctl flags, and Cloud Run completely abstracts host kernel configurations.
3
Evaluate VM provisioning types (Spot vs Standard VMs).
Spot VMs are eliminated because they can be preempted at any time with a 30-second warning, violating the requirement for uninterrupted 24/7 operation.
Spot VMs are designed strictly for fault-tolerant, stateless, or batch processing applications.
4
Select the optimal compute architecture.
GKE Standard with standard Compute Engine virtual machines provides container orchestration while permitting custom node OS kernel configuration and non-preemptible 24/7 availability.
GKE Standard node pools support custom sysctl settings through system configurations or daemonsets on standard VM instances.

Anahtar Kavram

Planning Compute Infrastructure and Evaluating OS/Workload Constraints
Bu soruyu puanla

Konular

Planning and Assessing Cloud Compute Solutions

Bu konudan daha fazla soru