Hiking to Harder Kulm, the top of Interlaken.

[Slides] [Project]

Kube-scheduler is a critical component to Kubernetes, responsible for placing the pod to the most suitable node. But how it works, can we customize it for advanced usage, what’s the best practice in large clusters. To answer these progressive questions, we’ll divide this session into two parts. If you’re a newbie to kube-scheduler, you may interest with the Intro part, if you’re a senior one, you can join our Deep Dive.

Parc des Princes.

New family member.

Stats

Coding

[Slides] [Project]

Compute demands for machine learning are growing rapidly nowadays. Ray, a unified computing framework, allows ML engineers to scale their workloads effortlessly without building complex computing infrastructures.

[Slides] [Project]

Kube-scheduler is a critical component to Kubernetes, responsible for placing the pod to the most suitable node. But how it works, can we customize it for advanced usage, what’s the best practice in large clusters. To answer these progressive questions, we’ll divide this session into two parts. If you’re a newbie to kube-scheduler, you may interest with the Intro part, if you’re a senior one, you can join our Deep Dive.

[Slides] [Project]

In this talk, Aldo and Kante will present the latest enhancements that SIG Scheduling recently promoted in Kubernetes, and the opportunities under discussion, to better support both services and batch type workloads in Kubernetes. We will discuss the recent improvements to scheduler performance that are allowing it to reach new scheduling throughput highs, better support for rolling updates in deployments while maintaining high availability, the new spec. SchedulingGates knob and how it allows external integrators, like dynamic quota managers, to control when pods should be considered for scheduling. We will also discuss the recent developments in sponsored projects, such as Kueue, scheduling plugins and the descheduler.

[Slides] [Project]

Batch, HPC and AI/ML clusters that serve multiple teams and researchers generally face the challenge of resource constraints. Cluster administrators want to control usage and maximize the utilization of resources. Kueue is a could-native job scheduler that works in combination with the default Kubernetes scheduler, the job controller and the cluster-autoscaler to provide a full batch system. Kueue implements job queueing, deciding when jobs should wait and when they should start, based on quotas and a hierarchy for sharing resources fairly among teams. Kueue works well in the cloud, where resources are heterogeneous, fungible and can be scaled up and down to save cost. In this talk, you will learn the key concepts of Kueue and how you can model your teams and resources to make your Kubernetes cluster a batch system.

Excited to earn the Kubernetes Contributor Award for the first time 🎉 !