Kubernetes Autoscaling: HPA, VPA, KEDA, and Cluster Autoscaler
Kubernetes provides a layered autoscaling stack that addresses horizontal replica scaling, vertical resource right-sizing, event-driven scale-to-zero, and node provisioning. This topic covers the HPA v2 algorithm and behavior block, including custom and external metrics via the Prometheus Adapter; VPA's three-component architecture, its four update modes, and the rule against combining VPA and HPA on the same resource metric; KEDA v2 and its ScaledObject/ScaledJob model that brings scale-to-zero to any workload through 60+ scalers; and node-level scaling through the Cluster Autoscaler (with its expander strategies and PDB awareness) and Karpenter (with NodePool-based just-in-time instance selection and consolidation).
Sign up free to read the full topic
More DevOps topics
- Kubernetes — Stuck Rollouts, Version Upgrades and Mistakes
- Kubernetes Admission Control
- Kubernetes AI and GPU Scheduling
- Kubernetes API Aggregation
- Kubernetes API Errors — 4xx and 5xx Explained
- Kubernetes Architecture
- Kubernetes Cluster Upgrades and Version Management
- Kubernetes Core Resources
- Kubernetes Custom Schedulers
- Kubernetes Deployment Strategies
- Kubernetes Descheduler
- Kubernetes Error Triage — Symptoms, Exit Codes and Cloud Equivalents
Start free on Cariara · Coding problems · System design studio