Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Professional Cloud Architect practice bank has 2,335 more like it. Try 10 free: instant score, no account.
An enterprise artificial intelligence team is deploying distributed large language model (LLM) fine-tuning jobs on Google Cloud. The architecture must satisfy the following technical and operational constraints:
Which compute provisioning strategy should the cloud architect recommend?
Future reservations combined with Committed Use Discounts (CUDs) allow organizations to contractually guarantee hardware capacity (such as TPU pods or high-end GPU clusters) for fixed continuous durations of one year or longer.
The organization were operating steady-state, continuous foundational model training pipelines or mission-critical 24/7 inference services that require guaranteed physical capacity and predictability over a multi-year horizon.
C2D machine types are compute-optimized VM instances powered by AMD EPYC processors designed for CPU-bound tasks like web servers, video encoding, and high-performance computing. Sole-tenant nodes provide dedicated physical compute servers exclusively hosting VMs from a single customer project for compliance and hardware isolation.
The architecture required isolated physical servers to satisfy strict BYOL (Bring Your Own License) per-core software licensing agreements or single-tenant data compliance rules for high-frequency CPU computing.
Spot VMs are excess Compute Engine capacity available at substantial cost discounts (typically 60% to 91% compared to on-demand pricing). They use preemptible quota and can be shut down by Google Cloud whenever capacity needs to be reclaimed for standard workloads.
The workload were a stateless batch processing pipeline, high-performance computing (HPC) simulation, or asynchronous rendering job with resilient checkpointing where random mid-job terminations cause minimal business impact.
Flex-start is a specialized consumption model powered by Google Cloud's Dynamic Workload Scheduler (DWS). It allows organizations to request high-performance accelerator infrastructure, such as TPUs and GPUs, for bounded durations of up to seven days without upfront long-term reservations or complex quota management.
Flex-start provides the optimal balance between high discount rates, guaranteed uninterrupted execution for up to 7 days, and dense accelerator topologies required for distributed ML training.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated