Examkin

Soru

Zorluk: Çok zorPlanning Compute Engine Resources and Machine Types

A biotechnology organization is designing the Compute Engine architecture for a genomics data platform featuring two distinct operational workloads:

1. Workload 1: A persistent, uninterrupted in-memory genomic reference database requiring at least 512 GB512\text{ GB} of RAM operating continuously 24/724/7 with strict availability SLAs.
2. Workload 2: A stateless batch processing pipeline that analyzes chunked genomic sequences asynchronously, can tolerate sudden instance termination with automated job re-queueing, and requires maximum cost reduction.

Which machine type family and provisioning strategy should you select to fulfill these requirements optimally?

  1. Provision Workload 1 on Memory-optimized (M-series) instances backed by 1-year or 3-year Committed Use Discounts (CUDs), and provision Workload 2 on Spot Virtual Machines using General-purpose instances.Cevap
  2. B
    Provision Workload 1 on Spot Virtual Machines using Memory-optimized instances to minimize database runtime costs, and provision Workload 2 on Compute Engine N2 standard instances relying on automatic Sustained Use Discounts.
  3. C
    Deploy Workload 1 to a GKE Autopilot cluster to eliminate node management overhead, and run Workload 2 on Cloud Functions for event-driven execution.
  4. D
    Deploy Workload 1 on Cloud Run with container memory limits scaled to 512 GB512\text{ GB}, and provision Workload 2 using N1 standard instances under a 3-year Committed Use Discount.

Cevap

Provision Workload 1 on Memory-optimized (M-series) instances backed by Committed Use Discounts (CUDs), and provision Workload 2 on Spot Virtual Machines using General-purpose instances.
The solution correctly identifies that high-memory, continuous workloads (≥512 GB\ge 512\text{ GB} RAM) require Compute Engine Memory-optimized machine types combined with Committed Use Discounts to secure maximum savings for 24/7 availability. Meanwhile, stateless and fault-tolerant batch workloads should use Spot VMs to leverage discount rates up to 91% without risking production data loss.

Adım Adım Çözüm

1
Analyze Workload 1 compute and memory specifications.
Workload 1 requires high RAM (≥512 GB\ge 512\text{ GB}) and 24/7 uninterruptible execution.
Memory-optimized (M-series) instances are designed for high memory-to-vCPU ratios (up to 30 GB per vCPU), and continuous 24/7 production usage warrants Committed Use Discounts (CUDs) for maximum cost savings without risking eviction.
2
Analyze Workload 2 operational tolerance and cost constraints.
Workload 2 is stateless, batch-oriented, fault-tolerant, and re-queues failed tasks.
Workloads that tolerate preemption are ideal candidates for Spot VMs, which offer deep discounts (60–91%) compared to standard pricing.
3
Synthesize the optimal Compute Engine resource configuration.
Match M-series + CUD for Workload 1, and Spot VMs for Workload 2.
This combination aligns performance, high availability, and financial optimization according to Google Cloud architectural best practices.

Anahtar Kavram

Compute Engine Machine Type Selection, Committed Use Discounts (CUDs), and Spot VM Workload Suitability
Bu soruyu puanla

Konular

Planning Compute Engine Resources and Machine Types

Bu konudan daha fazla soru