Medium1 markMultiple Choice
This question is part of a case study — click to read the full scenario(Case 06)

CASE STUDY: RetailMart

Overview: Global e-commerce, 5,000 employees. Legacy monolith on VMware, 20TB Oracle DB on-prem.
Business: Modernize to microservices, 100% uptime during Black Friday (10x traffic), real-time inventory sync, exit data center in 2 years.
Executives:

  • CEO: "Innovate faster to beat online-only competitors."
  • CFO: "End hardware CAPEX. Move to pure OPEX."
  • CTO: "Break monolith safely. Zero downtime during transition."
    Tech: Migrate off Oracle to open-source, containerize, secure hybrid connectivity during transition, automated scaling.
    Constraints: Zero downtime for storefront, heavy reliance on Oracle stored procedures, all hybrid traffic must be private/encrypted.

Which migration approach should you recommend to safely break down the monolith with zero downtime?

GCP PCA · Question 09 · Domain 5: Managing Implementation and Ensuring Solution and Operations Reliability

CASE STUDY: RetailMart

Overview: Global e-commerce, 5,000 employees. Legacy monolith on VMware, 20TB Oracle DB on-prem.
Business: Modernize to microservices, 100% uptime during Black Friday (10x traffic), real-time inventory sync, exit data center in 2 years.
Executives:

  • CEO: "Innovate faster to beat online-only competitors."
  • CFO: "End hardware CAPEX. Move to pure OPEX."
  • CTO: "Break monolith safely. Zero downtime during transition."
    Tech: Migrate off Oracle to open-source, containerize, secure hybrid connectivity during transition, automated scaling.
    Constraints: Zero downtime for storefront, heavy reliance on Oracle stored procedures, all hybrid traffic must be private/encrypted.

How should you optimize compute costs while ensuring the platform can handle Black Friday traffic spikes?

Answer options:

A.

Use Preemptible VMs for all GKE node pools to save 80% on compute costs.

B.

Purchase Committed Use Discounts (CUDs) for the baseline traffic and rely on GKE Cluster Autoscaler for the 10x Black Friday spikes.

C.

Provision enough on-demand Compute Engine instances to handle the 10x traffic year-round.

D.

Use Sustained Use Discounts (SUDs) exclusively, as they automatically apply to autoscaled resources.

How to approach this question

Separate the workload into 'baseline' and 'spiky'. Apply long-term discounts to the baseline and autoscaling to the spikes.

Full Answer

B.Purchase Committed Use Discounts (CUDs) for the baseline traffic and rely on GKE Cluster Autoscaler for the 10x Black Friday spikes.✓ Correct
Cost optimization in GCP relies on matching pricing models to workload profiles. RetailMart has a steady baseline of traffic year-round, which should be covered by 1-year or 3-year Committed Use Discounts (CUDs) for maximum savings. The 10x Black Friday spike is temporary, so GKE Cluster Autoscaler should dynamically provision on-demand nodes only when needed.

Common mistakes

Selecting Preemptible VMs (A) for a mission-critical production storefront. Spot/Preemptible VMs are for fault-tolerant batch jobs, not synchronous e-commerce APIs.

Practice the full GCP Professional Cloud Architect Practice Exam 4

50 questions · hints · full answers · grading

More questions from this exam