Collaborative Research: CNS Core: Small: Creating An Extensible Internet Through Interposition

International Computer Science Institute, Berkeley CA

Investigators

Abstract

In common parlance, the Internet is the set of applications and services that we can access from our phones, laptops, and other personal devices. However, in the networking research community, the term “the Internet” refers more specifically to the basic communication infrastructure on which these applications and services have been built. What many find surprising is that while the set of applications and services have changed dramatically over the years, the basic architecture of the Internet itself – that is, the arrangement of functionality in the communication infrastructure – has remained almost completely unchanged since it first took shape roughly forty years ago. This lack of change is not by choice; we know many ways the infrastructure could be improved to provide users with better security, privacy, and performance. Instead, the lack of change is because evolving such a large infrastructure is a difficult design challenge in itself. More specifically, any proposal for architectural change in the Internet must meet the twin challenges of: (i) how to deploy the new design without breaking the old, and (ii) how to create incentives so that all the major providers of the infrastructure benefit from this change (or else it is unlikely to happen). This research project aims to overcome these barriers with a design we call the Extensible Internet. If successful, this will not just be a one-time change, but will turn the Internet into a platform that can easily incorporate change on an ongoing basis. More technically, this work is built on the central insight that the original Internet architecture’s narrow waist of IP conflated two roles: that of tying all layer-2 (L2) networks together, and that of providing a service model to end hosts. The Extensible Internet separates these two roles by inserting an Interposition Layer between layers L3 and L4 that (i) is built on top of L3, and can use the current L3 without change, and (ii) provide the service model to end hosts. The entire focus of the Extensible Internet design is architecting this new layer so that an extensible set of services – ranging from secure attestation to improved privacy to multipoint delivery to various other delivery models – can be seamlessly deployed on top of this Interposition Layer. The research challenges include: how to structure the data path so that it can be both fully flexible and highly efficient; how to structure global services to be highly resilient; and how to feasibly support an evolving yet uniform service model (i.e., supported by all network providers). We will demonstrate the resulting design on various network testbeds, including NSF’s own FABRIC testbed. In addition, all software we create in this project will be publicly available and open source. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.