How Can Pod Start-up Be Accelerated on Nodes in Large Clusters?
Talkat KubeCon + CloudNativeCon + Open Source Summit China 2023 · on Thursday, 2023-09-28· at 14:45· 2层 会议室 1 | 2F Room 1·
summary
Explains concrete factors and optimization points that affect Pod startup time in large Kubernetes clusters, covering controller-manager timing for Pod creation, kube-controller-manager PV/PVC binding and webhooks, and scheduler concerns like GPU topology- and node-load-aware scheduling. Walks through kubelet-level issues (image pulling, sidecars, API QPS/burst, event-driven PLEG, throttling, disk/volume drivers, static CPU policies, container runtimes), GPU management (topology beyond NUMA, sharing, monitoring), data strategies (preloading, local vs distributed storage), and observability techniques to diagnose slow startups.
AI-written summary · about
related talks
08m⚡ Fast Restarts, Not Just Fast Starts: Accelerating Pod Recovery
34mPanel: Improving and Managing Kubernetes at Scale
30mMaximizing Launch Reliability: Controlled Lift-off for Reliable Application Startup on…
34mWhere's My Pod, Already? A Tale of Cold-Start Struggles on Kubernetes
29mProactive cluster autoscaling in Kubernetes
11mFast and Efficient Container Startup at the Edge via Dependency Scheduling
24mK8s Pod Scheduling - Deep Dive
15mDemo: Resource(Cost) optimization on planet scale Kubernetes clusters.