EKS Auto Mode & Karpenter

KubeCon + CloudNativeCon Europe 2026

Session Date: March 24, 2026 — London

What is EKS Auto Mode?

• Production-ready managed EKS experience
• Bundles Karpenter + Load Balancer Controller + managed node infrastructure
• Pricing: 10% premium per node over standard EKS
• Removes the need to manage node groups manually

Karpenter — Intelligent Node Management

• Drift Detection — automatically detects and reconciles configuration drift
• Consolidation — moves workloads to achieve optimal node utilization
• Spot Interruptions — 2-minute warning handled automatically; pods migrate seamlessly

Cost Optimization Features

Best Practices

• Pod Disruption Budgets (PDBs) — tell Karpenter how to safely drain pods
• Topology Spread Constraints — reduce blast radius across availability zones
• Node Disruption Budget — rules for when nodes can be optimized (maintenance windows, business hours)

Observability

• Managed Prometheus and Grafana plans expanding to eu-north
• Not all regions available yet — no firm timeline
• GroundCover — eBPF-based observability partner
  • Installs directly in-cluster
  • No sidecars, no code changes
  • Affordable alternative to vendor solutions

Key Takeaways

1. EKS Auto Mode is production-ready — simplifies node management at 10% cost premium
2. Karpenter's consolidation and drift detection are mature enough for production
3. Node Overlays enable fine-grained cost optimization
4. Use PDBs + Topology Spread + Node Disruption Budgets together for safe scaling
5. GroundCover offers a cost-effective eBPF observability option