Feedback Request: Workload Aware Scheduling & Autoscaling integration

[Paweł Kępka]

Hi all,

As I mentioned yesterday during the SIG Autoscaling meeting as part of the ongoing efforts to make kube-scheduler workload aware, we've been working on an proposal to better integrate scheduling and cluster autoscaling.

We've outlined a high-level architecture and the roles of three core integration APIs (Potential Capacity API, Pod Scheduling Preferences API, and Capacity Provisioning API) in the following document: Workload Aware-Scheduler Cluster Autoscaling.

These APIs aim to enable more transparent and cost-aware provisioning by giving the scheduler visibility into potential capacity and a mechanism to explicitly request it.

We would greatly appreciate your feedback, insights, and concerns on this proposal. Please feel free to add comments directly in the document.

Thanks!

[Jack Francis]

Thank you Paweł! Really excited to see these design discussions accelerate and materialize like this.

[Ellis Tarn]

Lots of great ideas in here. Left a few comments and looking forward to discussing more. 

--
You received this message because you are subscribed to the Google Groups "Autoscaling Kubernetes" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kubernetes-sig-auto...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/kubernetes-sig-autoscaling/815ee1b6-069d-414d-bcb5-eb1e66467202n%40googlegroups.com.