In Run:ai, what stops one team taking everything?
Why D is the answer
Project quotas
Quotas set each project's guaranteed share.
In detail
- Quotas set each project's guaranteed share.
- Quotas stop one team taking everything.
Use case
An illustrative scenario, not a real incident.
Each Run:ai project has a quota, which guarantees it a share.
See it in different ways
flowchart TD A["Departments"] --> B["Projects with quotas"] B --> C["Over quota when GPUs are free"] C --> D["Preempted when the owner needs them"]
mindmap
root((Sharing))
Run:ai
Project quotas
Over quota, then preempted
Slurm
Fairshare
MIG
GPU instances
Hardware isolationmindmap
root(("Cluster GPUs"))
n1["Project A"]
n2["Guaranteed quota"]
n3["Project B"]
n4["Guaranteed quota"]Key point: Project quotas guarantee shares.
Why the other options are wrong
- A. Larger Docker images
- Image size does not share GPUs.
- B. A larger shared node pool
- A larger pool does not divide it fairly.
- C. A shared password
- A password does not share GPUs.