A batch rendering job tolerates interruption and must cost as little as possible. Which compute do you choose?
Why D is the answer
Spot VMs
Spot VMs are the cheapest option for fault-tolerant work that can handle interruption.
In detail
- Spot VMs are the cheapest, and they can be interrupted.
- Save progress often.
Use case
An illustrative scenario, not a real incident.
A film studio renders thousands of frames. Interrupted frames just rerun, so Spot VMs cut the cost.
See it in different ways
flowchart TD
A{"Workload"} -->|"Short event handler"| B["Cloud Run functions"]
A -->|"Stateless container"| C["Cloud Run"]
A -->|"Complex container platform"| D["GKE"]
A -->|"Full VM control"| E["Compute Engine"]mindmap
root((Compute))
Cloud Run
Stateless containers
GKE
Kubernetes control
Compute Engine
Custom machine types
Spot VMs
VMware Engine
Existing VMwareflowchart TD
Q{"What happens on interruption?"}
Q -->|"It reruns safely"| A0["Spot VMs"]
Q -->|"Users are affected"| A1["Standard VMs"]Key point: Spot VMs suit interruptible work.
Why the other options are wrong
- A. Sole-tenant nodes
- Sole tenancy costs more.
- B. On-demand VMs
- On-demand costs more.
- C. Reserved VMs for three years
- A reservation is a commitment.