Never Go Limitless – On the Importance of ResourceQuotas at the Edge
summary
Edge clusters of one to three nodes have the control plane sharing resources with apps, and uncontrolled workloads can render the cluster unrecoverable without backups. The demo uses stress-ng and a simple fractal generator web app with autoscaling and ~400 and ~200 parallel request patterns to show CPU stays under nominal limits while memory exhaustion kills the system. Applying ResourceQuotas per namespace forces limits, keeps the cluster responsive, and prevents the reinstall scenario.
AI-written summary · about
related talks
34mA Huge Cluster or Multi-Clusters? Identifying the Bottleneck
31mLoad Testing Kubernetes: How to Optimize Your Cluster Resource Allocation in Production [I]
29mCapacity Constraints Unveiled: Navigating Cloud Scaling Realities
29mKubernetes Workload Resiliency in Action: Beyond Basics
47mOptimizing Kubernetes Resource Requests/Limits for Cost-Efficiency and Latency
40mLimits, Requests, QoS... Rethinking What You Thought You Knew About Scheduling in Kubernetes
11mLightning Talk: The Adaptive Edge: Balancing Cost Efficiency, Usability, and Performance Tradeoffs
24mFreedom of Kubernetes requires Chaos Engineering to shine in production