GCP-ACE - Planning and implementing a cloud solution (30% of the exam) - Section 2.1
Plan and implement compute resources, selecting among Compute Engine, Google Kubernetes Engine, Cloud Run, and Cloud Run functions for a workload, launching instances, choosing Compute Engine storage, creating autoscaled managed instance groups, using Spot VMs, and deploying GKE clusters and serverless applications that process Google Cloud events.
Match a workload to the right compute product, weighing Compute Engine, GKE, Cloud Run, and Cloud Run functions against control, portability, operational overhead, and scale-to-zero requirements. Create an autoscaled managed instance group from an instance template, choose zonal against regional Persistent Disk or Hyperdisk, use Spot VMs for interruptible work, deploy GKE Autopilot, regional, and private clusters, and wire serverless services to Pub/Sub, Cloud Storage, and Eventarc events.
Compute EngineGoogle Kubernetes EngineCloud RunCloud Run functionsManaged instance groupsEventarc
Practice question for this objective
lock_openFree samplePlanning and implementing a cloud solutionhard
A developer must run small pieces of code in direct response to Google Cloud events, such as an object being written to a Cloud Storage bucket, with the least infrastructure to manage. Which statement about Cloud Run functions is accurate for this event-driven need?
ACloud Run functions cannot respond to Cloud Storage events and must instead be invoked only by a scheduled Compute Engine cron job that polls the bucket.
BCloud Run functions require you to provision and patch a dedicated node pool in Google Kubernetes Engine before any event trigger will fire.
CCloud Run functions handle events but bill for a continuously running Compute Engine VM per function, so they carry the same operational overhead as a self-managed instance.
DCloud Run functions run your code in response to events on managed infrastructure, and can be triggered by Cloud Storage events delivered through Eventarc.check_circle Correct
Know that Cloud Run functions run event-driven code on managed infrastructure and consume Google Cloud events via Eventarc. Cloud Run functions are a managed serverless runtime that executes on demand in response to events, and Eventarc routes Google Cloud events such as Cloud Storage object changes to them without any customer-run polling or nodes.
Why A is wrong: This is tempting if you assume polling is required, but it is wrong: Cloud Run functions integrate with Eventarc to receive Cloud Storage events directly, so no polling VM is needed.
Why B is wrong: This confuses functions with GKE-hosted workloads; fully managed Cloud Run functions run on Google-managed infrastructure and need no customer-provisioned node pool to receive triggers.
Why C is wrong: This misstates the model: Cloud Run functions scale with events and do not bill for a per-function always-on VM, so their operational overhead is far lower than a self-managed instance.
Why D is correct: Cloud Run functions are the event-driven serverless option that runs code on Google-managed infrastructure and consumes events, including Cloud Storage object changes routed through Eventarc, matching the low-management event need.
Exam traps in Planning and implementing a cloud solution
Answers that look right on this material and are not. Each one is a distractor from a different question in the GCP-ACE bank for this domain.
Keep the zonal group but raise the autoscaler's maximum replica count so more Spot instances are requested during peaks.
errorWhy it is wrong: Requesting more instances in the same zone does not help when that zone is out of Spot capacity, since the shortage that causes the simultaneous reclamation is exactly what a higher ceiling cannot overcome.
Run a Compute Engine instance that polls the bucket every minute for new objects.
errorWhy it is wrong: A polling VM would work but adds a continuously billed instance and custom polling logic, which is more infrastructure than an event-driven trigger needs and introduces latency between upload and processing.
Run gcloud compute instance-templates update on api-tmpl-v1 to change the machine type to e2-standard-4, and let the group apply it to running instances.
errorWhy it is wrong: This is tempting because editing in place sounds simplest, but instance templates are immutable and there is no update subcommand, so the machine type of an existing template cannot be changed.