A platform team wants a list of Compute Engine VMs whose CPU and memory usage stays low enough that a smaller machine type would suit them, based on observed utilization rather than guesswork. Which source should they query?
Show answer and explanation
Answer: D
The Recommender service analyzes observed utilization metrics for Compute Engine instances and produces machine type recommendations that suggest a smaller or larger shape when usage consistently falls outside what the current type suits. These appear in the console and are available through the Recommender API. Billing exports show spend without utilization, Asset Inventory records resource metadata, and VPC Flow Logs describe network connections rather than CPU or memory use.
Why the other options are wrong
- A. A billing export shows what each VM costs by SKU and label. It can reveal expensive machines but carries no CPU or memory utilization, so it cannot tell whether a machine is oversized.
- B. Cloud Asset Inventory exports metadata such as machine type and configuration for each resource. It shows the current shape of a VM, not how heavily that VM is actually used.
- C. VPC Flow Logs record sampled network connections between endpoints. Traffic patterns say nothing about CPU or memory load, so they do not support a downsizing decision.