GKE 10 years and SIG Networking, With Antonio Ojea

In this episode of the Kubernetes Podcast from Google, hosts Abdel Sigiwar and Kaslyn Fields interview Antonio Ojea, a software engineer at Google and a core maintainer of Kubernetes. They discuss the evolution of networking from hardware-based to software-defined networks in Kubernetes, addressing the challenges of integrating traditional networking with Kubernetes' API-driven approach. Antonio provides insights on treating clusters as autonomous systems, defining ingress and egress points, and standardizing network policies. The conversation covers specific issues like egress traffic, IP address whitelisting, and the impact of AI/ML workloads on networking requirements. They also explore the Gateway API project, multi-service CIDR, and the complexities of achieving quality of service for networking in Kubernetes, including the challenges posed by AI and large model downloads. The discussion further extends into the future of AI in networking, the use of RDMA for high-performance computing, and security concerns related to MCP servers.