Current State of MPA Project and Integration of Network/I/O Metrics

[Parthiba-Hazra]

Hi team,

I’m reaching out with a couple of questions regarding the Multi-dimensional Pod Autoscaler (MPA) project. One of our current project aims to combine the strengths of HPA and VPA, and we’re also exploring the possibility of incorporating additional dimensions—specifically network and I/O metrics—to better handle scenarios where these resources become the bottleneck.

Given that the MPA project, I have the following inquiries:

Current Status:
What is the current state of the MPA project? Are there any early designs, roadmaps, or milestones that we could review?

Network and I/O Metrics:
Have there been any discussions or plans to integrate network latency and I/O metrics into the autoscaling recommendations? In our context, where our platform supports arbitrary virtual Kubernetes clusters, such metrics could greatly improve scaling decisions and overall performance.

Thanks in advance for your time and feedback!