Making The Case To Rebuild The Internet From Scratch
How can the Internet run more efficiently? A growing number of experts say so-called "band-aid" fixes are insufficient -- and the entire system of communication needs to be rebooted.
PARIS – Videos that take hours to load on a computer; companies fending off hacking attempts; inboxes filled with spam; users that are a little too anonymous for some governments; a network that is too U.S.-centric for Beijing…
These are just some of the reasons that researchers, companies and countries want to change the Internet.
Some companies have already developed a “layer” over the Internet for their own purposes. Skype has high quality demands for voice and video communication and Akamai offers to speed up their clients’ flow through its own network.
In fact, several research teams around the world are working on this issue. In the U.S. there is the National Science Foundation’s FIND project, as well as the GENI and RINA projects. Europe has FIRE, funded by Brussels. China and Japan are in the race too. They all have opposing interests but they do agree on one thing: A network designed for a few hundred scientists cannot be expected to meet the same requirements as a vortex sucking in more than 2.4 billion users -- especially when these users connect from multiple devices, do online commerce, and perform surgical operations remotely.
Some researchers actually want to start over with a “clean slate.” John Day, director of the RINA project in Boston, is one of the advocates of this solution. They want to replace the 30-year-old TCP/IP protocol developed by Internet co-founders Vint Cerf and Bob Kahn.
The Internet is no longer a network of networks, but a single big network. In the 1980s, while many companies were creating private networks but the U.S. government released its free, open-source TCP/IP protocol and froze the initiatives that could have led to a thriving Internet -- an Internet that would have developed slower but would have been more secure, according to Day.
Day wants to fix a fundamental design flaw: On the Internet, IP addresses confuse identity and location. In other words, we give addresses to machines, not actual people. What may seem like a minor detail isn’t. For instance, it makes the network less “redundant” -- whereas the Internet was in fact created to provide this type of redundancy -- side roads that can be taken in case the main communication road is blocked.
Today, whenever a company registers with two different Internet Service Providers for security or budget reasons, it will be assigned two separate IP address ranges. If they were merged under one single denomination, we would have a better chance of contacting them. This IP address problem results in all sorts of constraints: allocation of network resources, mobility management…
“If we start over with a clean slate, we could decide to identify contents or services instead of machines, it would be a complete paradigm shift,” says Kave Salamatian, professor of computer science at the French University of Savoie. This would be in the best interest of companies like Google. Whereas they are currently forced to negotiate with telecom providers to locate content servers in their facilities, this could grant them independence. Needless to say ISPs are less enthusiastic.
“Today on the Internet, everything is changing: on the top layer, the apps; in the infrastructure, the 4G and fiber-optic technology. The only thing that hasn’t changed is the IP protocol,” says Salamatian.
On the other hand, he says, we can still continue to apply "band-aids" on the IP. Stephane Bortzmeyer, R&D engineer at France’s domain name administration (AFNIC), believes that the band-aids will prevail, because the “clean slate” underestimate the weight of the existing system.
“We already have so many problems with the deployment of IPv6 – the solution to the shortage of IP addresses in the original Internet – that I can’t even imagine how they could build what would not be a new Internet, but a whole new network!” he says.
Band-aids already exist, such as the BCP 38 standard developed to stop IP address spoofing. But ISPs don’t use it, he explains: “They don’t want to pay for it since it is other companies that will benefit from it.”