You need to deploy an application in Google Cloud using serverless technology. You want to test a new version of the application with a small percentage of production traffic. What should you do?

Question

Options

• ADeploy the application to Cloud Run. Use gradual rollouts for traffic splitting.
• BDeploy the application to Google Kubernetes Engine. Use Anthos Service Mash for traffic
• CDeploy the application to Cloud Functions. Specify the version number in the functions name.
• DDeploy the application to App Engine. For each new version, create a new service.

• A
  78% (52)
• B
  3% (2)
• C
  6% (4)
• D
  13% (9)

Explanation

Cloud Run is the correct serverless choice here because it natively supports traffic splitting between revisions, allowing you to route a specific percentage (e.g., 5%) of live traffic to a new version - exactly what the question asks for.

Why the distractors are wrong:

• B (GKE + Anthos Service Mesh) is not serverless - GKE requires managing Kubernetes clusters, and Anthos adds significant complexity beyond what's needed.
• C (Cloud Functions) doesn't support traffic splitting by version name; function versioning isn't a built-in traffic management feature, and naming conventions alone won't split traffic.
• D (App Engine) actually does support traffic splitting, but the action described - creating a new service per version - is incorrect. In App Engine, traffic splitting is done between versions within a service, not across services.

Memory tip: Think "Cloud Run → Run a percentage of traffic to a new revision." Cloud Run's revision-based model with percentage-based traffic controls is the canonical GCP serverless answer for canary deployments and A/B testing.

No community discussion yet for this question.