K8s Resource Management: autoscaling

[kevinintel]

No description provided.

[jfding]

The resource management of Kubernetes for OPEA workloads is a very big problem, we need to break down to a few specific requirements in the future, according to the feedbacks from developer and landing customers. This issue can be a placeholder currently, to collect all related feedbacks.

[jfding]

+Malini as well to work on this ticket.

[daisy-ycguo]

I don't think it is a valid issue.
@kevinintel @jfding @mkbhanda what's your option ? how to handle it?

[mkbhanda]

Let us make this a feature request, a longer-term endeavor. It is not a bug. Also OK to close it for now if @jfding agrees.

[endomorphosis]

the term "resource management" is too vague for me to provide any meaningful input beyond the changes to the manifests so that multiple containers of the same service are not staged on the same machine.

[kevinintel]

It’s a feature of roadmap.

[endomorphosis]

Okay, but i dont know what the design goals behind "resource management" are, e.g. autoscaling, s3 model storage, etc, its a vague term

[kevinintel]

#144 : autoscaling
V1.0 #144
Q1/25: opea-project/GenAIEval#58
Please sync with Kaajas, Samu

[kevinintel]

#386