Navigating the Next Frontier: The Annual eBPF and Cilium Gathering for AI-Scale Kubernetes
As artificial intelligence and graphics processing unit (GPU) workloads become standard in cloud-native environments, traditional Kubernetes networking architectures are hitting their limits. To address these emerging challenges, a specialized gathering dedicated to the eBPF ecosystem is taking place. This event, running on November 9, 2026, in Salt Lake City, serves as a critical meeting point for engineers tackling high-scale data paths and runtime security. It offers a focused space to address the specific bottlenecks that arise when AI and GPU workloads push container orchestration past its traditional boundaries.
The Unified eBPF Foundation
At the core of this discussion is the idea of a single, unified technological foundation. The ecosystem relies on eBPF to power everything from network routing and observability to runtime security enforcement. By focusing on this shared base, the gathering ensures that conversations remain deeply technical and directly applicable to production environments. Attendees will explore how a consolidated data path allows for seamless integration between network policies, traffic monitoring, and threat detection, eliminating the need for fragmented solutions.
Scaling for AI and High-Performance Computing
AI and GPU-driven applications demand extreme network performance and specialized hardware acceleration. The event will dive into the friction points that arise when migrating traditional Kubernetes clusters to support these workloads. Topics will include the intricacies of RDMA policy enforcement, strategies for implementing IPv6-only cluster meshes, and methods for tracing packet loss in latency-sensitive inference pipelines. Real-world case studies from major tech infrastructure providers will highlight the practical challenges of large-scale data path migrations and how teams successfully navigated them.
Kernel-Level Security in a Fast-Paced Threat Landscape
Security cannot be an afterthought, especially as automated vulnerability discovery accelerates the release of new CVEs. This gathering emphasizes runtime security enforcement directly within the kernel, bypassing the need for application-side modifications. Discussions will cover enforcing mutual authentication between pods without altering application code, alongside deploying granular runtime policies that detect and respond to suspicious behavior in real-time. These capabilities ensure that AI clusters maintain strict isolation without sacrificing the performance gains of specialized hardware.
Event Overview and Target Audience
Scheduled as a co-located event during the broader North American cloud-native conference, this gathering is tailored for platform engineers, infrastructure specialists, and open-source contributors. Whether you are currently operating these tools at scale or evaluating them for future adoption, the agenda offers practical, production-tested insights. Attendees will leave with actionable strategies for common yet complex issues, such as debugging high-performance network traffic and securing isolated compute environments.
FAQ
Q: Why is a dedicated event necessary for this technology?
A: Because the challenges faced when scaling AI and GPU workloads require highly specialized knowledge. General conferences cover too broad a range of topics, whereas this gathering focuses exclusively on the deep technical nuances of the eBPF-based networking and security stack.
Q: What types of problems will attendees learn to solve?
A: Participants will gain insights into running Kubernetes entirely on IPv6, enforcing network policies on high-speed RDMA traffic, tracing packet loss in low-latency environments, and debugging inference workloads without impacting performance.
Q: Is prior experience with these tools required to attend?
A: A foundational understanding of Kubernetes networking is recommended for most sessions. However, newcomers can prepare using available educational resources about the underlying technology and its applications. Advanced sessions may require familiarity with multi-cluster architectures and specific data path migrations.
Q: How does the security component integrate with the networking discussions?
A: Security is treated as an extension of the networking layer. By using kernel-level enforcement, teams can authenticate pod traffic and detect threats without adding sidecar proxies, ensuring that security measures do not introduce bottlenecks in high-performance environments.
Q: Who are the typical speakers and attendees?
A: The audience includes platform engineers running these systems in production, maintainers of the open-source projects, and infrastructure teams from major tech companies who have navigated large-scale migrations and can share their real-world experiences.
Conclusion
As cloud-native infrastructure evolves to meet the relentless demands of artificial intelligence and high-performance computing, the tools and strategies used to manage them must evolve as well. This annual gathering provides the essential forum for engineers to exchange real-world solutions, confront the limits of traditional networking, and collaborate on the future of the eBPF ecosystem. By focusing on practical, production-ready insights, the event equips attendees with the knowledge needed to build faster, more secure, and highly scalable Kubernetes environments.
Thank you for reading
