* Function is linear between points. Resource utilization is defined as one minus ratio of total amount of resource requested by pods on node and node's capacity (scaled to 100).

- DRA: scheduling pods is up to 16x faster, depending on the scenario. Scheduling throughput depends a lot on cluster utilization. It is higher for lightly loaded clusters with free resources and gets lower when the cluster utilization increases. ([#127277](https://github.com/kubernetes/kubernetes/pull/127277), [@pohly](https://github.com/pohly)) [SIG API Machinery, Apps, Architecture, Auth, Etcd, Instrumentation, Node, Scheduling...

  - Remove redundant logging around calculating node utilization
  - Add configurable `--network` and `--rm` flags for docker in `Makefile`
  - Subtract DaemonSet pods' requests from node allocatable in the denominator while computing node utilization
  - Include taints by condition when determining if a node is unready/still starting

  - `priority_level_seat_count_watermarks` is removed.
  - `priority_level_request_count_samples` is replaced with `priority_level_request_utilization`, which samples every nanosecond rather than every millisecond; the old metric conveyed utilization despite its name.
  - `priority_level_request_count_watermarks` is removed.

  - Remove redundant logging around calculating node utilization
  - Add configurable `--network` and `--rm` flags for docker in `Makefile`
  - Subtract DaemonSet pods' requests from node allocatable in the denominator while computing node utilization
  - Include taints by condition when determining if a node is unready/still starting

- Fixed patch/update operations on multi-version custom resources ([#70087](https://github.com/kubernetes/kubernetes/pull/70087), [@liggitt](https://github.com/liggitt))
- Reduced memory utilization of admission webhook metrics by removing resource related labels. ([#69895](https://github.com/kubernetes/kubernetes/pull/69895), [@jpbetz](https://github.com/jpbetz))

   ([#123492](https://github.com/kubernetes/kubernetes/pull/123492), [@carlory](https://github.com/carlory))
- Added metric name along with the utilization information when running `kubectl get hpa`. ([#122804](https://github.com/kubernetes/kubernetes/pull/122804), [@sreeram-venkitesh](https://github.com/sreeram-venkitesh))

- Fix scoring for `NodeResourcesMostAllocated` and `NodeResourcesBalancedAllocation` plugins when nodes have containers with no requests. This was leaving to under-utilization of small nodes. ([#102925](https://github.com/kubernetes/kubernetes/pull/102925), [@alculquicondor](https://github.com/alculquicondor))

pod priority and pod preemption. These features allow mixing various types of workloads in a single cluster, and help reach
higher levels of resource utilization and availability.
These features are in alpha. SIG Scheduling also improved the internal APIs for scheduling and made them easier for other components and external schedulers to use.

* kubelet: resolved hang/timeout issues when running large numbers of pods with unique configmap/secret references ([#74755](https://github.com/kubernetes/kubernetes/pull/74755), [@liggitt](https://github.com/liggitt))
* Reduce memory utilization of admission webhook metrics by removing resource related labels. ([#69895](https://github.com/kubernetes/kubernetes/pull/69895), [@jpbetz](https://github.com/jpbetz))