SIG-Scheduling Intro and Deep Dive

Tips for optimizing Kubernetes scheduling throughput and customization

• Adjust the percentage of nodes to score based on cluster size
• Iterate a few nodes instead of all of them in the default preemption plugin
• Use the prefer nominated node feature to save time in scheduling
• Customize the parallelism parameter to optimize performance
• Model customer resources as extended resources to schedule GPU-intensive workloads
• Use multi-profile scheduling to specify different scheduling policies for different workloads
• Simplify scheduled configuration with a multi-point plugin

Suppose a user has a cluster of 2000 nodes and specifies the percentage as 10. The filter phase will only score 200 nodes, saving time and resources. Additionally, the prefer nominated node feature can assign a node to a workload without going through the regular scanning flow, further optimizing scheduling throughput.

SIG Scheduling is responsible for kube-scheduler and its related subprojects such as the scheduling-plugins and descheduler. These projects share the common goal of enabling users and developers to ensure that pods are assigned to the appropriate nodes based on various scheduling criteria. We also endeavor to ease customizing and extending scheduler for advanced usage. This talk will introduce attendees to SIG-Scheduling with an overview of the current goals and ongoing work within the SIG, as well as opportunities for new contributors to get involved.