Is the Internet doomed to fail?

Network World - Sounds like a crazy question, in this era of Facebook, Twitter and a "digital millennial" generation that's grown up never not knowing the Internet.

But there are worrying signs that the Internet's architecture may not be able to scale effectively much longer. I've written previously about the crisis in access capacity: Demand for access bandwidth is growing exponentially, while provider investment is growing linearly. The lines cross -- demand exceeds capacity -- sometime around 2012. And the solution isn't for providers (or even the government) to simply invest more money in access -- because Internet access is a break-even business at best, and a money-losing proposition in many cases.

That's just the tip of the iceberg. Noted journalist John Markoff has written about systemic security flaws in the 'Net. And more recently, the folks at the IETF have finally admitted that IPv6 deployment -- intended to address addressing and routing scalability in the next-generation Internet -- has failed. (That exact word -- failure -- was used in an IETF presentation last October).

Is it possible there's a fundamental flaw within the Internet itself, that the architecture is simply not sustainable going forward? And more importantly, should the Internet be sustainable? Do we really need an Internet?

Sounds like a crazy question -- but it's not. Both content and service providers have seriously questioned the need for an Internet in recent months. Google and Amazon seem to be taking the perspective that so long as users can access their content, it doesn't matter if they can access anyone else's -- so they're constructing overlay networks for their own content. And a major telco recently asked me -- in all seriousness -- why it should bother maintaining the 'Net.

In other words, both content and service providers seem to be advocating a return to the "walled garden" model, where if users want an application, they need to buy the network to get to it.

My view? We absolutely, positively need the Internet. The real value in the 'Net is that it enables anyone, anywhere, to build out an application that everyone can access. That kind of openness is what's driven the amazing productivity increase that's occurred in the 1990s and 2000s. Returning to walled gardens simply won't cut it.

That brings me to the second question: Are there fundamental architectural flaws to the 'Net? Yes, there are. I've written about those in previous columns, and will continue to track them in future ones.

But the good news is that they can be fixed, and a cadre of researchers are working on this as we speak. For more information, check out sessions on the topic at the upcoming FutureNet conference (May 4-7 in Boston). Full disclosure: My Nemertes Research colleague, Irwin Lazar, is the content director for the event -- but the great content comes from the researchers and technologists who will be presenting.

Reprinted with permission from

For more information about enterprise networking, go to NetworkWorld.com
Story copyright 2009 Network World, Inc. All rights reserved.