I don't know when it started, but it became visible a couple of years ago. Now it may be the thing that makes that much-more-publicized concept moot. What is it? The notion that perhaps we need to rethink completely the thing we call the Internet.
MIT's David D. Clark sounded an alarm about the Internet's problems back in about 2004. His major concern is security, an Internet issue that's become clear to everyone. Clark feels the Internet has an intrinsic security problem that should be solved in a way that's more fundamental than gluing security on the edges. My own projections seem to confirm his worry: If security costs continue to rise at their current rate and bandwidth costs continue to fall at their current rate, we'll reach a point in the early part of the next decade where we will spend more on security than on the Internet.
Clark also tells a funny story about a fellow professor who remarked to him that the Internet had left out the most important protocol. "What was that?" Clark asked, eager to work on fixing the problem. The professor replied, "The one that routes the money!" That points out the second problem; the Internet well may be the only example of "bill and keep" in any retail system in the world.
What would happen if airlines worldwide flew passengers according to a model where the first airline that passengers board sells them a ticket for the first part of the trip and keeps all the money, but passengers are able to make connections to other airlines and fly anywhere? You can bet that on those connecting airlines you'd get minimal service at best, and not too many would stay in business. Settlement is a part of every wholesale or retail supply chain and every telecom service -- except the Internet. As far back as the mid-1990s, ISPs themselves were saying that without settlement there'd never be Internet QoS because the full chain of providers involved in Internet flows couldn't be compensated for playing their part.
Clark isn't the only one who's saying that something new is needed; the topic even is reaching the mainstream print media. The Philadelphia Inquirer ran an article late last month titled "The clean slate way to redo the Internet." It quoted Rutgers University professor Dipankar Raychaudhuri as saying that the Internet "was designed for completely different assumptions" and that "it's sort of a miracle that it continues to work well today."
The Inquirer article raises broader concerns, all rooted in the fact that the Internet's design reflects the constraints of the early 1970s. Computing in that period was based on monolithic mainframes and dumb terminals. Technology and application requirements have moved beyond that simple beginning. Has time stood still in the network world? If it hasn't, the Internet's mission also must have changed, along with the technology on which it is based.
There's more to this than just media hype and speeches, too. The National Science Foundation in the United States, as well as a number of projects in the European Union, is examining new architectures for the Internet. The IPsphere Forum has taken up the issue of how to support all the possible models of billing and settlement over IP networks and networks built of hybrid IT, Ethernet and optical technologies. How fast these processes will move forward is hard to say, but it is clear there is growing momentum to examine many of the Internet's basic properties and designs.
As the Inquirer article points out, it won't be as easy to redesign the Internet as it was to design it originally. There was no commercial use back in the 1970s, no users to scream about changes in their services and no vendors with turf to protect. There are legitimate questions about whether it would be best to do the clean-slate changes some propose or take a more adaptive approach, as the IPsphere Forum suggests. It's clear that the "Internet" of the future probably will have to support the current browsers and other consumer devices and tools, so the changes are likely to come deeper in the core, where they don't have an immediate impact on users.
Some signs of those changes may be emerging already. BT, whose 21CN is the most ambitious convergence project ever undertaken, has recently told its suppliers that they will have to support Provider Backbone Transport (PBT), an Ethernet-based trunking architecture, in addition to IP. Many believe that PBT is a threat to IP/MPLS, which has been the linchpin of router-vendor convergence strategies. The most significant thing about PBT, however, may be that it could be the first widely deployed non-IP technology to carry IP traffic.
The Internet is one of the most important technical innovations of modern times, and changes in how it works will have far-reaching consequences. We're probably on the cusp of a period of great changes. Eventually, those changes probably will make things better, but there surely will be stumbles and roadblocks for vendors -- and users -- to overcome.