csperkins.org

Panel on the Future of Internet Transport at IFIP Networking 2020

I was pleased to join a panel discussion at the IFIP Networking 2020 Workshop on the Future of Internet Transport today, along with Jana Iyengar and Alessandro Ghedini, and moderated by Gorry Fairhurst. The slides I used to frame my position are available, and if you're registered for the conference, a recording of the discussion is also available.

The description of the panel was as follows: "A new design for the transport system might introduce platform- and protocol-independent transport layer APIs, such as Apple’s Network.framework, or the IETF TAPS API. This has the potential to enable innovation in the design of transport protocols. What are the challenges facing Internet transports as user patterns and applications evolve at an ever-increasing speed? Re-designing the interface presented to applications, could enable changes to the way applications interact with the transport system, or the way they describe their use of the network. What information is helpful for an application to tell the transport system? Design and configuration of network equipment has often assumed relatively little knowledge of the way a transport protocol operates, but a patch-work of network-layer methods have ossified around the transport wire image. Middleboxes, in particular, have been deployed to augment transports to meet the demands of particular network segments or to map applications to appropriate network services. The growing volume of encrypted traffic demands a fresh approach. What information is helpful for an transport to tell the network system? What service model is needed for different applications? Increased data rates and processing capabilities present opportunities for greater intelligence in network devices, including new functions and software-based features. What are the opportunities for innovation at the network layer to assist the transport layer? If new approaches succeeded in addressing the ossification problem, what intelligence can the transport expect from the network layer? How should today’s research be evaluated against the expected changes in network traffic, and a more rapid pace of evolution above and below the network interface? Are there transition paths that can lead to successful long-term evolution of Internet transports? This is a panel session that will start with opening remarks by the panel, and then discuss thoughts on the topics brought to the table. Come prepared to listen, think and contribute your thoughts or your own questions!"