AWS re:Invent 2022 - Kubernetes Keynote Review

The Kubernetes Story at re:Invent 2022

AWS re:Invent 2022 was a turning point for Kubernetes on AWS. The announcements were not splashy product launches. They were infrastructure maturation signals. AWS stopped pitching Kubernetes as a novelty and started treating it as core infrastructure — the same way they treat EC2, S3, and RDS.

The keynote reinforced a clear message: AWS is committed to EKS as the primary container orchestration platform. Not ECS. Not Fargate alone. EKS. Every significant container announcement centered on making Kubernetes operations smoother, more integrated, and more enterprise-ready.

EKS Anywhere: Kubernetes Beyond the Cloud

EKS Anywhere was a major focus. AWS expanded its on-premises Kubernetes story with improved support for bare-metal deployments, VMware vSphere integration, and Snowball Edge environments. The message was unmistakable: Kubernetes is not just a cloud tool. It is the universal compute layer.

For enterprises with hybrid requirements — data sovereignty, latency-sensitive workloads, or regulated environments that cannot run entirely in the cloud — EKS Anywhere provides a single operational model. Same EKS API, same CLI tools, same deployment patterns, whether the cluster runs in us-east-1 or in a private data center.

The practical value depends on operational maturity. EKS Anywhere shifts control plane management to your team. If you have a strong platform engineering team, this is freedom. If you do not, it is a liability. Assess your team's Kubernetes operational depth before committing to on-premises EKS.

The EKS Add-Ons Ecosystem

AWS expanded the EKS add-ons catalog significantly. VPC CNI, CoreDNS, kube-proxy, and the EBS CSI driver were already available as managed add-ons. re:Invent 2022 signaled a broader strategy: make the most common Kubernetes operational tools installable and upgradable through a single API.

This matters because add-on lifecycle management is one of the biggest operational headaches in Kubernetes. When you upgrade EKS from 1.24 to 1.25, you also need to upgrade your CNI plugin, your CSI drivers, your ingress controller, and your observability agents. Managed add-ons reduce that coordination burden.

The limitation is selection. AWS curates the add-on catalog conservatively. If you need a specific version of Istio, a particular Prometheus operator configuration, or a non-AWS ingress controller, you are still managing those yourself. Managed add-ons complement a platform engineering approach. They do not replace it.

AWS Container Strategy: Reading Between the Lines

The broader container strategy at re:Invent 2022 told a clear story. AWS is consolidating around three container services: EKS for Kubernetes workloads, ECS for teams that want AWS-native orchestration without Kubernetes complexity, and Fargate for serverless container execution.

The investments were disproportionately in EKS. More add-ons. Better observability integration with CloudWatch Container Insights. Tighter IAM integration with IRSA (IAM Roles for Service Accounts) and the newer EKS Pod Identity. AWS is betting that most enterprise container workloads will run on Kubernetes, and they are building the ecosystem to support it.

Fargate's role shifted subtly. It is no longer positioned as an ECS alternative. It is positioned as a compute option for both ECS and EKS — a way to run pods without managing EC2 nodes. For burst workloads and workloads with strict isolation requirements, Fargate on EKS is a compelling option.

What This Means for Enterprise Kubernetes Adopters

If you are building on AWS and evaluating container strategies, the re:Invent 2022 announcements gave a clear signal: invest in EKS. AWS is investing more in EKS tooling, integration, and enterprise features than in any other container service. Our Kubernetes consulting team has seen this play out consistently with our AWS clients.

The EKS ecosystem is mature enough for production workloads across every major industry. Managed node groups reduce EC2 lifecycle management. Karpenter (which launched at a previous re:Invent) provides intelligent, fast autoscaling. Bottlerocket, AWS's container-optimized OS, reduces the host attack surface.

The operational bar is still higher than managed services like Lambda or Fargate. Kubernetes requires investment in platform engineering, observability, and team training. But for workloads that need the flexibility, portability, and ecosystem of Kubernetes, EKS on AWS is a strong foundation.

THNKBIG's Take: Container Services Maturation

We have been deploying EKS clusters since EKS launched in 2018. The maturation we saw at re:Invent 2022 validated what we have observed on the ground: EKS has moved from "early adopter" to "default choice" for AWS container workloads.

The areas that still need work are multi-cluster management, cost visibility at the namespace and pod level, and GitOps-native cluster lifecycle management. AWS has made progress on each, but organizations running more than a handful of EKS clusters still need third-party tooling or significant custom automation.

If you are planning your Kubernetes strategy on AWS, the decisions you make around cluster architecture, networking, security, and CI/CD will determine your operational costs for years. Our AWS partnership and hands-on EKS experience mean we help you get those foundational decisions right from the start.

Plan Your Kubernetes Strategy on AWS

Whether you are migrating to EKS, optimizing an existing cluster, or evaluating EKS Anywhere for hybrid workloads, our engineers bring production-tested expertise. Talk to an engineer and get a Kubernetes architecture review tailored to your AWS environment.

Key Takeaways

• AWS re:Invent 2022 declared the decade of containers over and the decade of serverless and AI/ML infrastructure beginning — while simultaneously announcing significant EKS feature improvements.
• The Kubernetes-related keynote content highlighted AWS's investment in simplifying EKS operations: EKS Anywhere improvements, Managed Node Groups updates, and Karpenter reaching general availability.
• Karpenter's GA was the most operationally significant announcement for EKS users — replacing the Cluster Autoscaler with a faster, more cost-efficient alternative.

Karpenter GA: The Operational Impact

Karpenter replaces the Kubernetes Cluster Autoscaler for AWS EKS users. The behavioral difference is significant. Cluster Autoscaler evaluates pending pods and triggers node group scaling operations — a process that can take 3-5 minutes. Karpenter watches pending pods directly and provisions EC2 instances in 60-90 seconds, selecting the optimal instance type for the pending workload's requirements.

Karpenter also deduplicates aggressively: it provisions exactly the instance type and size that fits pending pods, avoiding the over-provisioning that Cluster Autoscaler's fixed node group sizes create. Teams migrating from Cluster Autoscaler to Karpenter typically see 15-25% reduction in EC2 spend within the first month, combined with faster scale-out during traffic spikes.

EKS Anywhere and Hybrid Infrastructure

AWS's investment in EKS Anywhere — the on-premise distribution of EKS — addresses the significant number of enterprises that need consistent Kubernetes tooling across cloud and on-premise environments. EKS Anywhere uses the same EKS Distro (the Kubernetes distribution AWS uses in managed EKS) and connects to AWS for cluster management, logging, and observability while keeping workloads on-premise.

For Texas and California enterprises with regulatory requirements that limit cloud data residency but want consistent operations tooling, EKS Anywhere provides a practical hybrid path. THNKBIG's Kubernetes consulting practice includes EKS, EKS Anywhere, and multi-cluster implementations. Talk to our team.