Goal

The descheduler will be enhanced to balance VMs accross nodes, according to pressure (for overutilized clusters) and cpu (for undeutilized clusters). Configuring the decheduler will be done with a specific virt profile. The CNV UI will be enhanced to do the configuration graphically (CNV-52012).

The following changes will be implemented

• implement taint controller in Descheduler Operator - CNV-53978
• implement deviation thresholds for even better balancing - CNV-57713
• implement "double taints" for convergence when combined with deviation thresholds - CNV-53978
• implement cpu utilization based balancing for balancing of underutilized clusters - CNV-57900
• implement a dedicated profile in the descheduler operator for the LaB use-case (q: different profiles for pressure vs utilization?) - CNV-59897
• implement automatic switching between pressure and utilization based LaB - CNV-57900

not all of this work is in this epic, but all of this work is what we need to push fwd for CNV.

User Stories

• As a cluster administrator, I want balance my underutilized cluster, so that all nodes have the same spare capacity (cpu utilization based balancing)
• another user story

Non-Requirements

• List of things not included in this epic, to alleviate any doubt raised during the grooming process.

Notes

• Any additional details or decisions made/needed