Google ACE Planning and Configuring a Cloud Solution Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Use a VM with GPUs for faster processing

  Why it's wrong here

  Adding GPUs would only help if the batch job were doing massively parallel compute such as machine learning or rendering; for typical data processing, the bottleneck is CPU, I/O, or scalability, not GPU throughput. GPUs also dramatically increase per-hour cost and require quota approval, so they are an inefficient way to reduce nightly processing costs. Moreover, using GPUs on a preemptible VM is possible but still more expensive than a preemptible CPU-only VM, and it does not address the underlying interruptible nature of the workload.

• ✗

  Use a VM with local SSD for temporary storage

  Why it's wrong here

  Local SSDs are physically attached to the host and provide extremely low latency, but their data is ephemeral: it is permanently lost if the VM is terminated, preempted, or if the host undergoes maintenance. A nightly batch job needs durable storage for input, intermediate checkpoints, and output, so local SSDs would not provide a safe place for results and would need to be re-populated on every run. Replacing them with persistent disk or Cloud Storage is more appropriate for batch processing, and this option does not lower compute cost at all.

• ✗

  Use a standard VM and commit to a 1-year commitment

  Why it's wrong here

  A one-year committed use contract is designed for steady-state, always-on workloads where you agree to pay for a fixed amount of vCPU/memory for the full term. A nightly batch job that only runs for minutes or hours would be underusing that committed capacity, making the per-run cost unnecessarily high. Committed use discounts also do not apply to preemptible VMs, and you would be locked into a long-term expense that offers no benefit for a short, interruptible job.

• ✓

  Use a preemptible VM

  Why this is correct

  Preemptible VMs cost up to 60–80% less than standard on-demand VMs and are explicitly designed for fault-tolerant, batch workloads that can be interrupted. Compute Engine can terminate a preemptible VM at any time, but it will always run for at least 30 seconds, and the job should be coded to handle early termination by persisting progress to durable storage. Because this nightly batch job is by nature interruptible and short-lived, preemptible VMs are the cost-optimal choice and align with Google's best practices for batch processing.

Go deeper

Related to this question