Skip to content

Conversation

rodrigorfk
Copy link

@rodrigorfk rodrigorfk commented May 13, 2025

Issue #, if available: This is actually a feature request, so I haven't created an issue for it before

Description of changes:

Some times, depending how your workloads are structured in terms of resources requests, mostly if the amount of cpu and memory requested do not follow an appropriated ratio, you can end up with a cluster allocation like the one in the following example:

  1. CPU allocation
    image

  2. Memory Allocation
    image

As you can see, the allocation on each dimension (CPU and Memory) is relatively low, and it is hard to tell if Karpenter for instance, is doing a good job on consolidating and choosing the right instance type for bin-packing the workloads.

However, if eks-node-viewer would offer a flag to enable a normalized view of resources, meaning, if when watching for CPU resources, the memory allocation on a given node is higher than the cpu, the viewer would normalize the CPU allocation on the node to match the memory allocation, increasing it to show the node is actually higher allocated on CPU as well, and vice versa.

For the example above, when this logic is applied, you can actually see the cluster is 99.6% allocated, clearly showing that Karpenter is doing a good job and there is pretty much no wasted resources from a normalized point of view.
image

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

… the higher allocated dimension among CPU and Memory

Signed-off-by: Rodrigo Fior Kuntzer <[email protected]>
@rodrigorfk rodrigorfk requested a review from a team as a code owner May 13, 2025 14:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant