Explanation:
For non-interruptible workloads, use on-demand resources scheduled to run only during needed time (e.g., via Cloud Scheduler + VM start/stop or Cloud Run Jobs). Spot can be preempted, always-on clusters/services waste cost for batch.
Ultimate access to all questions.
You need to run a containerized job that processes a queue for ~3 hours each night and must not be interrupted. Cheapest suitable compute?
A
Spot VMs
B
A standard Compute Engine VM (or Cloud Run job) started on schedule and stopped after
C
A GKE Autopilot cluster running 24/7
D
Always-on Cloud Run service with min instances = 5
No comments yet.