I've seen the future of the Internet, and it's recursive -- in more ways than one.
There's an excellent new book out called Patterns in Network Architecture: A Return to Fundamentals (Prentice Hall) by the inimitable John Day, who was one of the original developers of Internet architecture. It's at least as compulsively readable and technically detailed as my other favorite network architecture book (Computer Networks, a Systems Approach, by Larry Peterson and Bruce Davie, now in its fourth edition).
But Patterns in Network Architecture is far more wide-ranging and ambitious in scope. A computer scientist and historian, Day digs into what I like to call '"the story behind the story'" -- not just how a protocol works, but how it came to be adopted, and why other approaches were (rightly or wrongly) discarded. Best of all, he doesn't neglect the human factors that come into play in the process -- and he's not afraid to tell it like it is, often in ways that are both erudite and laugh-out-loud funny. As my fellow reviewer and noted Internet researcher Dr. Jon Crowcroft observed, you'll often see Wittgenstein, Dave Clark, Confucius and Dr Seuss all cited to make a point.
The result is quite possibly the most panoramic technical and social review of the Internet's development ever written -- along with a detailed set of prescriptions and recommendations for its future. Some key takeaways:
-- The Internet, despite its enormous scalability and commercial success, is fundamentally a prototype, desperately in need of a major overhaul if it is to be relied on as a critical component of 21st century infrastructure.
-- Many of the technical challenges facing the Internet today are the results of design decisions taken in the 1970s that were made on the basis of politics, ego and emotion rather than a dispassionate assessment of their effectiveness.
-- The current mindset governing the vast majority of Internet-related research is overly conservative. As Day writes, "It begins to seem that people are the keepers of some flame and dare not tamper with what has been entrusted to them -- the classic behavior of a 'second generation'." This mindset is a primary factor that's holding back needed architectural innovation
-- Finally, and most importantly, Day proposes a recursive model for interprocess communication (IPC) that elegantly and effectively addresses many of the intractable architectural challenges, including the address limitations that are driving carriers to a disruptive forklift IPv6 upgrade, multihoming and mobility constraints, and route table scalability. The recursive IPC model blows up a few long-held assumptions (such as the significance of layers and the unique importance of global addressing) and dramatically simplifying congestion control, call management and QoS.
And speaking of recursion, implicit throughout the book is Day's keen awareness that those who cannot remember the past are condemned to repeat it. Part of his goal is clearly to ensure that past mistakes aren't repeated. That's ambitious and deeply worthwhile -- as is the book. I strongly recommend it as a thought-provoking and entertaining read.