Maximizing Workload's Performance With Smarter Runtimes

Explore innovative approaches to maximizing container workload performance and resource utilization in this conference talk from KubeCon + CloudNativeCon North America 2021. Delve into the challenges of mitigating noisy neighbors in containerized environments and discover new extension points for containerd and CRI-O. Learn about the Node Resource Interface (NRI) proposal and the evolution of dynamic resource usage optimizations in the CRI-Resource-Manager project. Gain insights into handling heterogeneous CPU resources, multi-tiered memory, caches, memory bandwidth, and block I/O usage. Understand how leveraging various metrics and hints provided by the Linux kernel can lead to improved workload performance and dynamic hardware resource utilization optimizations.

Intro
Performance Optimization
CPU Architecture
Linux Architecture
Caches
Kubernetes
Pluggable Policies
Node Resource Interface
Summary
Outro