In a significant boost for enterprises blending cloud and on-premises Kubernetes environments, Amazon Web Services has rolled out expanded support for Cilium networking in its Elastic Kubernetes Service (EKS) hybrid nodes. This update, detailed in an AWS announcement dated August 2025, allows users to leverage Cilium’s advanced features like eBPF-based networking and security across hybrid setups, streamlining operations for workloads that span AWS cloud instances and on-premises hardware.

Cilium, an open-source project under the Cloud Native Computing Foundation, has gained traction for its high-performance networking capabilities without relying on traditional iptables. In the context of EKS hybrid nodes—introduced to enable running Kubernetes pods on non-AWS infrastructure while managing them through EKS—this expanded support addresses previous limitations. Previously, hybrid nodes required mixing CNIs like AWS VPC CNI for EC2 instances and Cilium or Calico for on-premises nodes, leading to complexity in routing and policy enforcement.

Simplifying Hybrid Networking Challenges

The new support enables Cilium to operate as the sole CNI across the entire cluster, a move hailed in a Medium post by Jean-Francois Nadeau from June 2025. Nadeau’s analysis highlights benefits such as simplified routing in tunnel mode using VXLAN encapsulation, which minimizes dependencies on underlying physical networks. This is particularly useful for organizations with diverse infrastructure, reducing the overhead of maintaining multiple networking plugins.

According to AWS documentation on configuring CNIs for hybrid nodes, Cilium now supports overlay modes with masquerading disabled, ensuring pod CIDRs are routable without NAT complications. This facilitates seamless communication between cloud-based and hybrid pods, enhancing scalability for AI/ML workloads that demand low-latency interconnects.

Practical Deployment Insights and Performance Gains

A practical guide on AWS re:Post from April 2025 walks through deploying Cilium on EKS hybrid clusters, covering static and BGP routing configurations. It emphasizes load balancing options, which are crucial for high-availability setups. Insiders note that this integration could cut deployment times by up to 30%, based on community benchmarks shared in recent X posts from AWS enthusiasts.

Further, a March 2025 AWS re:Post article unpacks networking options, comparing Cilium’s eBPF efficiency against alternatives. With EKS now supporting up to 100,000 nodes per cluster—as announced in a July 2025 AWS blog post—Cilium’s role in hybrid environments becomes even more pivotal for ultra-scale AI training, potentially handling 1.6 million accelerators.

Security and Observability Enhancements

Cilium’s Hubble observability tool integrates natively, providing real-time visibility into network flows, which is invaluable for debugging hybrid clusters. A May 2025 guide on Palette’s documentation for importing EKS clusters and enabling hybrid mode underscores the ease of configuring Cilium, aligning with AWS’s push for commercial support on core features.

Industry experts, echoing sentiments from X discussions around AWS’s August 2025 posts, praise this for bolstering security through identity-aware policies. However, AWS cautions that advanced customizations may require in-house expertise or third-party support, as outlined in their hybrid nodes CNI docs.

Future Implications for Enterprise Adoption

This update positions EKS as a more flexible platform for hybrid cloud strategies, potentially accelerating migrations from legacy on-premises Kubernetes. A Medium article by Amit Gupta from July 2025 details Cilium’s support for Elastic Fabric Adapter (EFA) in EKS, enhancing performance for NCCL tests in AI workloads.

As enterprises grapple with data sovereignty and cost optimization, Cilium’s expanded role could redefine hybrid Kubernetes management. With AWS’s recent scaling announcements, including support for massive node counts reported in WebProNews four days ago, the combination promises robust, efficient infrastructures for next-gen applications.