Why should you bother about cluster autoscaling- Karpenter

on YouTube ↗21m

" summary

Karpenter claims to provision nodes directly by calling cloud compute APIs (example AWS EC2 Fleet) instead of expanding ASGs, and it picks the smallest set of instance types that fit pending pods (m5.xlarge instead of matching t3.mediums in the demo). It prioritizes spot instances and falls back to on‑demand, uses ttlSecondsUntilExpired and node‑empty TTLs to delete nodes, and binds pods during provisioning so images can prepull — all aimed at faster, groupless scaling and lower retry latency.

AI-written summary · about

" chapters (12)

1. 0:00Karpenter intro
2. 2:52Why cluster autoscaling matters
3. 3:36How Cluster Autoscaler works
4. 8:38Karpenter intro
5. 9:41How Karpenter provisions nodes
6. 13:33Karpenter with provisioners
7. 15:19Spot and on-demand mix
8. 16:43Karpenter operations
9. 17:46Single vs multiple provisioners
10. 19:31Topology spread constraints
11. 19:53Karpenter advantages
12. 21:20Q&A

" related talks

• 36mGroupless Autoscaling with KarpenterEllis Tarn, Prateek Gogia · KubeCon + CloudNativeCon Europe 2021
• 27mScaling K8s Clusters Without Losing Your Mind or MoneyYael Grossman · Conf42 Cloud Native 2023
• 01h 01mJust-in-time Nodes for Any AWS EKS Cluster: KarpenterSamuel Baruffi · Conf42 Cloud Native 2023
• 22mKarpenter Efficient scaling of Kubernetes clustersViktor Velmich · AWS Summit Tel Aviv 2023
• 27mHarness the power of Karpenter to scale, optimize & upgrade K8sChakkree Tipsupa · Conf42 Kube Native 2024
• 22mUse Karpenter to scale, optimize & upgrade K8sChakkree Tipsupa · SREday London 2024
• 19mJust in time Worker Nodes with KarpenterRohini Gaonkar · KCD Sri Lanka 2022
• 18mKarpenter : Your trusted Companion for Kubernetes ScalingToshal · KCD Hyderabad 2024