The next computing paradigm: P2P?

The growth of VOIP is difficult to miss these days - I can't think of one major software or internet entity that is not doing something in this area. Most recently the launch of Google Talk last week triggered volumes of opinions across many technology outlets - in fact this blog posts track-backs to a number of them. I fully expect before long we will see furious activity in the plugin segment as Google continues to build a consciousness with their growing band of loyal developers.

From my position, I think the growth of peer-to-peer VOIP is growing evidence of an emerging architecture that presents a challenge to the traditional tiered architectures that have formed the basis of software design for some time.

I think we've been here before, but in a very different context. Consider how traditional US domestic carriers have historically relied on the central hubs as their basis for their networks. They now find themselves out maneuvered by a new model which champions a rapidly growing, low-cost, but most importantly point-to-point route networks. It has become a successful model particularly for Southwest Airlines and RyanAir in Europe.

It's possible that we are about to see a similar challenge to infrastructure that has built up as a result of millions of engineering hours. How large scale installations match the nimble and vastly more adaptable model that peer-to-peer offers? One has only to look at the differences between what I personally consider the most predominant VOIP services today - Vonage and Skype.

Vonage relies on a large scale, traditional PSTN style switching network to enable its service -- Skype simply requires a broadband connection and a simple software installation thus allowing it to grow organically without the need for huge investments in infrastructure (obviously, Skype's central user database in an exception in this case). The differences in the quality of service they provide is stark - Skype provides a near in-room experience with it's sublime voice quality -- and your calls are secure. Quite simply the peer-to-peer model makes this possible.

Let me connect this back to how this will pose a challenge to tiered architectures. Increasingly innovation around how to use established technologies that have serviced us well in tiered architectures will be applied to the peer-to-peer model. Just imagine how query based technologies such as SQL and more likely XQuery will play a roll in this evolution in this new brand of software. A brave new world awaits for those that revolutionize query based applications based on peer-to-peer relationships forever breaking out of the mold of tiered models...