-
Epic
-
Resolution: Done
-
Critical
-
None
-
None
-
None
-
Resource manager recommendation for AI clouds
-
L
-
False
-
-
False
-
Not Selected
-
Proposed
-
Proposed
-
In Progress
-
RHOSSTRAT-512 - RHEL AI on RHOSO - Telemetry
-
Proposed
-
rhos-workloads-vaf
-
Proposed
-
0% To Do, 0% In Progress, 100% Done
-
-
More specific user stories must be defined for individual use cases of resource management.
This should be done in conjunction with work by DFG:Workloads, especially in case of requirements for resource management.
Following topics must be addressed for the epic resolution:
- scheduling
- GPU quotas
- Oversubscription prevention
- Autoscaling for inference
Some of these, specifically the autoscaling, may already be resolved by ShiftStack. Result of the epic should be creation of user stories.
