Overview
Explore a comprehensive conference talk on SIG Scalability in Kubernetes, presented by Wojciech Tyczyński and Marcel Zięba from Google. Dive into the various efforts of SIG Scalability, including defining scalability for Kubernetes, driving improvements, and implementing infrastructure for scalability testing. Learn about the Kubernetes Scalability Envelope, cluster happiness metrics, Scalability SLOs, API Call Latency SLOs, and scalability limits. Discover tools like ClusterLoader2 and Kubemark for cluster simulation, as well as observability and debuggability features such as Peridash and profiling. Gain insights into scalability test infrastructure, notable 1.22 scalability regressions, and recent scalability improvements. Engage with the presenters during the Q&A session to understand how SIG Scalability can better collaborate with the community and provide input on the roadmap for future work.
Syllabus
Intro
What do we do?
Kas Scalability Envelope
Defining happiness of cluster
Scalability SLO
API Call Latency SLO
Scalability definition
Scalability Limits
ClusterLoader2
Kubemark - Cluster Simulation
Observability & Debuggability - Peridash
Observability & Debuggability - Profiling
Scalability Tests infrastructure
Notable 1.22 Scalability Regressions
Recent Scalability Improvements
Taught by
CNCF [Cloud Native Computing Foundation]