Capacity Scheduling for Elastic Resource Sharing in Kubernetes
CNCF [Cloud Native Computing Foundation] via YouTube
Overview
Explore a collaborative presentation on capacity scheduling for elastic resource sharing in Kubernetes. Learn how Qingcan Wang from Alibaba and Yuan Chen from Apple address the limitations of fixed resource quotas in multi-tenant environments. Discover a Kubernetes enhancement that enables dynamic resource sharing within and across teams, improving cluster utilization. Gain insights into hierarchical resource groups with guaranteed and maximum resources, supporting various resource types including CPU, memory, disk, and GPU. Understand how this feature, similar to YARN's capacity scheduler, integrates seamlessly into Kubernetes as plugins and its implementation in large-scale production clusters like Alibaba Cloud. Follow the presentation's structure, covering the background, Kubernetes resource management, resource guarantees, extensions, nested structures, job queue architecture, and a live demonstration.
Syllabus
Introduction
Background
How does Kubernetes manage resources
Resource Guarantee
example
extension
nested structure
job queue
job queue architecture
demo
current status
Collaboration
Questions
Taught by
CNCF [Cloud Native Computing Foundation]