Using the same adapter for the Vertical Pod Autoscaler (VPA)

[andriyfomenko]

Expected Behavior

The same kind of annotations-based "magic" to work for VPA, in the same way as it is exposed for HPA

Specifications

It's just an idea, no spec [yet]

[jonathanbeber]

Hello, @andriyfomenko, thanks for opening the discussion. Although, I think the best place to discuss that is on the VPA project. This project is just an implementation of a custom metrics API server.

I see recently it was part of a discussion in SIG autoscaling and, we can follow this issue for that: https://github.com/kubernetes/autoscaler/issues/4135

Maybe, it's worth explaining, in that issue, how you would make use of custom metrics to scale your workloads. It would be beneficial for the discussion that's starting. Mention what kind of metrics and how you would map them to CPU and memory vertical scaling.