New resource: ScaleGroup

[hoerup]

Proposal

I propose a new CR called ScaleGroup where you can set a max limit of all ScaledObject or ScaledJob that refers to that ScaleGroup

Use-Case

Lets say that I have 20 different ScaledObject that each can scale from 0 to 1000 and that is usually ok because they never trigger at the same time. But if worst case happens and they all trigger at the same time they could overflow my cluster if it's fixed scale or use unwanted funds if ClusterAutoScale is enabled.

Therefore it would be great if i could group some of the scaled objects and define a maximum of pods that are allowed to scale within that group

The group CR should of course be configurable on what to do if limits is reached. should it just stop scale
Should it try to even out the scaled object via some formula

Is this a feature you are interested in implementing yourself?

No

Anything else?

No response

[SpiritZhou]

Could you provide a more detailed ScaleGroup definition, such as a ScaleGroup YAML with possible properties, so that we can discuss it more precisely instead of just conceptually?