GCP Associate Cloud Engineer Practice Question 3: Spot VM Batch Workload
Google Cloud Associate Cloud Engineer · Test 1 · question 3 of 10
A media company runs a nightly video transcoding job on Compute Engine. The job is fault tolerant: workers checkpoint progress to Cloud Storage every few minutes and resume from the last checkpoint if a worker disappears. A single run frequently takes about 30 hours of wall-clock time, and the company wants the lowest possible compute cost.
Which compute configuration best fits these requirements?
Show answer and explanation
Answer:
Spot VMs are excess Compute Engine capacity sold at discounts of up to 91% off the on-demand price. They can be preempted at any time, which is acceptable here because the workers checkpoint and resume. Critically, Spot VMs have no maximum runtime unless you set one yourself, so a run lasting roughly 30 hours is fine.
Option B is the trap. Preemptible VMs are the older version of the same idea and carry a hard 24-hour maximum runtime; Compute Engine always terminates them within 24 hours even if they are never preempted for capacity. A 30-hour run would be cut off every time, so Spot VMs, not preemptible VMs, are the right choice.
Option C removes the interruption risk but at a much higher price. Committed use discounts are typically in the range of a few tens of percent and lock you into a one- or three-year commitment for a workload that only runs at night, a poor fit for bursty batch work that tolerates preemption.
Option D solves an entirely different problem. Sole-tenant nodes give you physically dedicated hardware for licensing or compliance isolation and cost *more* than shared on-demand instances; they do nothing for cost optimization here.
Reference: Spot VMs