Evolving Multi-Network in Kubernetes: From Pod Spec to Dynamic Resource Allocation
UA2.114 (Baudoux) | Day 2 | 15:25 - 15:45 | Speakers: Miguel Duarte, Doug Smith
Abstract
At last year's FOSDEM, Doug and Miguel presented about upcoming efforts to improve networking APIs in Kubernetes, such as the work done by the SIG-Network Multinetworking Working Group, and efforts to evolve CNI, and... Most of them have been blown out of the water.
How Networking in Kubernetes will work is changing directions rapidly, driven in part by the evolution of dynamic resource allocation (DRA). We're here to draw the map of where the community is going, and to help guide you through the naval mines that might lurk beneath the surface for your implementations and cluster configurations.
Key points we'll cover: - Recap of the multi-network challenge and last year's proposed solution - Introduction to Dynamic Resource Allocation (DRA) and its applicability to multi-network scenarios - Deep dive into KEP-4817, which proposes enriching the Resource Claim Status with standardized network interface data - Comparison of the DRA approach with previous solutions like Multus CNI - Practical implications for developers and operators looking to implement multi-network solutions - Discussion on the future of multi-network in Kubernetes and potential next steps
We'll also share insights on the community collaboration process, highlighting how feedback and consensus-building have shaped the current direction. We'll unpack what went right, what got torpedoed, and what it means for the future for Kubernetes native multi-networking, and how to move the several adjacent communities (CNI, NPWG, device management, kubernetes sig-network) forward.
Join us to learn about the latest developments in Kubernetes networking and participate in shaping the future of multi-network support in cloud-native environments.
Attachments
Speakers
Links
External Links
Notice: The placeholder video image is licensed under CC BY-SA 4.0. The original image can be found hereChanges made to the image are: Cropped the image to a new ratio, part of the image was cut off.
