Faster nodes, smarter scaling: What’s new inside Amazon Elastic Kubernetes Service (Amazon EKS) Auto Mode

This article details performance and scalability improvements shipped to Amazon EKS Auto Mode across four infrastructure pillars.

• Node boot time reduced 39% (13 seconds faster) through startup detection optimization
• Karpenter delivers 43% faster scale-out and up to 69% faster consolidation with 30% more cluster capacity
• Node-local DNS provides sub-millisecond resolution without cluster-wide bottlenecks
• zram memory compression protects system components from OOM events on smaller instances
• Faster container image pulls via increased registry QPS, NVMe optimization, and SOCI parallel pull
• Topology-aware EBS volume scheduling supports mixed Auto Mode and traditional node group clusters
• Separate pod subnets and security groups enable enterprise networking configurations
• IPv4 egress in IPv6 clusters and DNS-based network policies improve connectivity options
• All improvements deploy automatically with no configuration changes required

EKS Auto Mode now delivers faster, more efficient, and more resilient infrastructure for production Kubernetes workloads without requiring manual upgrades or configuration changes.