Intro: SIG Autoscaling
summary
Overview of Kubernetes autoscaling components and current status with actionable details: Horizontal Pod Autoscaler uses metrics-server for CPU and supports custom metrics via adapters so HPA can target non-CPU signals; recommended CPU target is 70–80% CPU target; Vertical Pod Autoscaler runs a recommender/updater/admission pipeline and offers dry‑run, recommended and full‑control modes; Cluster Autoscaler adds or removes nodes when pods are unschedulable and supports GCP, AWS, Azure and Alibaba; HPA GA on CPU, HPA custom metrics beta, VPA alpha moving to beta, cluster‑autoscaler GA.
AI-written summary · about
chapters (12)
- 0:00Start
- 1:03How Much Resources Do You Need in Your Cluster
- 4:29How To Maintain the Decent Load in Your Application
- 8:33Custom Metrics to Horizontal Path Autoscaler
- 9:14How To Use It
- 12:04Vertical Path Autoscaler
- 14:27When To Use It
- 15:27Can You Mix Horizontal Pot Autoscaler and Vertical Pot Autoscaler at the Same Time
- 17:50Scheduling
- 21:33How To Use Cross Delta Scyl
- 22:54Status Report
- 29:23Is There a Plan To Like Support Prometheus Custom Metrics Natively
