New High-Performance Features in Red Hat OpenShift

Explore five new high-performance features in Red Hat OpenShift in this 45-minute conference talk. Discover how Kubernetes and Red Hat OpenShift have evolved to support diverse and complex applications through demonstrations of hugepages, device plugins, extended resources, GPU-accelerated machine learning TensorFlow workloads, and CPU assignment with tuned profiles. Learn about Red Hat's involvement in establishing the Kubernetes Resource Management Working Group and gain insights into running performance-sensitive workloads on OpenShift. Get an overview of resource management in OpenShift and a glimpse into the future of upstream Kubernetes. Delve into topics such as snowflake workloads, hardware accelerators, compute resource management, quality of service, cluster topology, and challenges with CPU management.

Syllabus

Introduction
Snowflake workload
Highperformance workloads
Latency sensitive
Gigantor
Hardware accelerators
Work notes
Lets not give up
Roadmap
Userfacing API
Compute Resource Management
Quality of Service
Cluster topology
Node config changes
Challenges with CPU
CPU bound limits
Static CPU management
Demo
Huge Pages
Consuming Huge Pages
Huge Pages Demo
Extended Resources
GPU Enablement
Concrete Example
YouTube Example
GPU Demo
SIS Controls
Summary