Explore network-aware scheduling in Kubernetes through this conference talk by José Santos from Ghent University. Learn about a new framework designed to optimize latency-sensitive applications by considering network factors in addition to traditional resource metrics. Discover how the framework incorporates two controllers (AppGroup and NetworkTopology) and three scheduling plugins (TopologicalSort, NodeNetworkCostFit, and NetworkMinCost) to ensure bandwidth reservations and minimize end-to-end application latency. Gain insights into the benefits of this approach for applications like databases and multi-tier web services. Watch a demonstration showcasing the framework's advantages and understand the challenges and lessons learned in implementing network-aware scheduling in Kubernetes environments.
Overview
Syllabus
Introduction
Background
Framework
Latency
Framework overview
Bandwidth
Example
Network topology cr
Topological sorting
Latency optimization
Filtering and scoring
Cluster topology
Filtering function
Scoring function
Demo
Deploying Controller
Networkaware Performance
Challenges Lessons Learned
Thank You
Questions
Taught by
CNCF [Cloud Native Computing Foundation]
