Network Bandwidth-Aware Kubernetes Cluster - Enhancing Scheduler for Network Resources
Explore a 36-minute conference talk that delves into the challenges of network bandwidth oversubscription in Kubernetes clusters and proposes innovative solutions. Learn how the current Kubernetes scheduler's focus on CPU and memory allocation falls short in addressing bandwidth contention issues for sensitive applications. Discover an approach to extend the default Kubernetes scheduler to an external service, enabling improved placement decisions based on network resources. Gain insights into a method for ensuring consistent network bandwidth for critical applications in virtualized infrastructures like vSphere or OpenStack. Understand the importance of reserving network bandwidth for worker nodes across the entire cluster and incorporating this information into scheduling algorithms.