Containers
Karpenter v1.15: GPU-aware provisioning, smarter node repair, and more
Karpenter v1.15 brings hardware-aware provisioning for NVIDIA GPU and EFA workloads through Dynamic Resource Allocation, volume-limit-aware scheduling, and Nitro Enclaves support, plus node repair as voluntary disruption with reboot-based recovery, terminate-first disruption, pod deletion cost management, and CEL expressions in kubelet configuration.
Your guide to containers at AWS re:Invent 2026
AWS re:Invent 2026 returns to Las Vegas, November 30 to December 4. This guide to the container track groups Amazon ECS, Amazon EKS, AWS Fargate, and Amazon ECR sessions by the kind of work you do, from running agents in production to platform engineering, inference, security, scale, and resilience, so you can start building your schedule.
Use Amazon ECR global authentication token across AWS Regions
Amazon ECR now supports global authentication tokens: request one token from AWS Sign-In and use it to push and pull images across AWS Regions in the same partition, with a lifetime from 5 minutes to 12 hours. This post shows how to request a global token and use it with Docker in two Regions.
Running GitHub Actions runners with Amazon EKS Auto Mode
Learn how to run self-hosted GitHub Actions runners on Amazon EKS Auto Mode. Actions Runner Controller with EKS Auto Mode, EC2 Spot Instances, and AWS Graviton scales runners to match CI/CD demand and scales to zero when idle, cutting idle compute costs and manual node operations.
High-quality cost-effective AI image generation on Amazon ECS
Learn how to run Z-Image-Turbo, a 6-billion-parameter diffusion transformer, on AWS Inferentia2 and Amazon ECS to generate high-quality images at a fraction of a cent each, with auto-scaling, arbitrary resolution support, and a complete CI/CD pipeline.
How Salesforce efficiently right-sizes Kubernetes workloads at scale on Amazon EKS
Salesforce operates one of the largest multi-tenant Kubernetes platforms on Amazon EKS, spanning 1,400+ clusters. See how its team combined the Vertical Pod Autoscaler and Horizontal Pod Autoscaler into a multi-dimensional autoscaling strategy that right-sizes workloads at scale and drove more than $5 million in infrastructure savings.
Automatically recover from GPU failures with Amazon ECS Managed Instances
Amazon ECS Managed Instances provides built-in GPU health monitoring and automatic repair powered by NVIDIA DCGM. This post shows how GPU auto-repair detects critical GPU hardware failures and replaces impaired instances with no code changes, and how to observe and safely validate the workflow using DCGM fault injection on a dual-tier inference pipeline.
AWS Nitro Sandboxes: Hardware-enforced workload isolation on Amazon EKS
AWS Nitro Sandboxes bring hardware-enforced, per-pod isolation to Amazon EKS. Every sandboxed pod runs in its own lightweight virtual machine with the isolation of a dedicated Amazon EC2 instance and the simplicity of a Kubernetes pod. Learn how it works and how to turn it on with a one-line change to your pod spec.
Deep dive: Auto scaling response times on Amazon ECS
Amazon ECS now publishes high-resolution 20-second CPU and memory utilization metrics for target tracking, so service auto scaling detects and responds to load changes faster. This post shares AWS benchmarking results across AWS Fargate and ECS Managed Instances and shows how to turn the feature on.
Measuring developer experience on an Amazon EKS developer platform
Most platform teams can’t prove their internal developer platform is improving software delivery. This post shows how to automatically collect the four DORA metrics (deployment frequency, lead time for changes, change failure rate, and failed deployment recovery time) on an Amazon EKS platform using Argo CD, Argo Rollouts, Apache DevLake, and Grafana.









