The End of Internet Architecture

Architecture. There’s a lot of it about. Do we need it? Recent years have seen considerable publishing activity in the area of “internet architecture”. This paper steps back and asks a more radical question: is an internet architecture a good thing at all? Are we at a point in the development of distributed communications systems where the concept should be replaced by a different way of dividing up the design space? This is not a paper about what the Right internetwork architecture should look like, but rather whether the very idea of a network architecture at this point in history is a help or hindrance in moving communication technology (and research in particular) forward. We first examine critically what the role of the internet architecture is today. We argue that instead of acting as useful guide for network practitioners of all kinds (as it has in the past), the principle function the architecture performs these days is to keep the fields of “networking” and “distributed systems” separate, to the detriment of both. Put simply, the internet architecture, and more broadly the concept of a network architecture, is now in the way. At the same time, trends in networking and computational hardware, and in particular in the kinds of testbeds available to researchers to validate their ideas, have made it both feasible and compelling to do research that finesses network architecture as an issue completely, and concentrates on the broader problem of building, deploying, and operating large-scale distributed applications. Fortunately, this does not render research into “internet architecture” irrelevant, but it does call for a respinning of many of the ideas in a different context. This paper concludes by examining the new research opportunities in the area, and how they relate to tradition challenges in “architecture”.