Net overhaul long overdue

A networking expert claims the internet as it is currently constructed is nearing collapse

THINK THE internet was created to avoid a nuclear attack? That it is a flat–structured democracy where no one has any formal control? That it is a wonderful example of endless innovation? That it runs on carefully crafted protocols which are regularly updated to deal with new problems?

Allow John Day to disabuse you of those notions. An adjunct professor at Boston University and an expert at networking with a wicked sense of humour, Day is adamant that the internet as it is currently constructed is nearing the point of collapse, a crisis that will not be solved by releasing more internet addresses or making adjustments to the networking protocol that underlies its structure, called IP, or Internet Protocol.

Day was involved from the 1970s in developing communications protocols used in the internet as well as its predecessor, Arpanet. And he is none too happy at how it has turned out, with many of today’s problems based on poor early decisions about addressing (how people would find things on the internet) and routing (how people would get to those things).

“The internet is living on Moore’s Law and Band-Aids,” he says, launching into a list of its problems during a seminar in Dublin last week organised by Waterford Institute of Technology’s Telecommunications Software and Systems Group (TSSG).

“And much of what people believe about the internet is a myth.”

Far from being a hotbed of innovation, he insists “the internet has been stagnant since 1975”, with little done to correct problems he says its initial designers knew were there from early on. His ultimate insult? “If the internet were an operating system, it would be DOS,” he says, referring to the famously sloppy Microsoft operating system that came before Windows.

It also isn’t a democracy but “more like the Communist Party”, with decisions on design and implementation as subject to the politics of organisations and personalities.

“Most problems start with IP (internet protocol) and the lack of a complete addressing structure,” he says. IP addresses, sequences of numbers that are often said to indicate a destination on the internet, actually indicate the final bit of the route to that destination. It’s as if the number of a house actually referred only to the street it is on, rather than the house itself.

The problem goes back to old telecommunications protocols based on the copper phone wires running into homes back in the 1950s. “Traditionally, people always named the wire,” Day says; originally, the last four digits of phone numbers in the United States identified the wire running into the house.

On the internet, this has led to a confusing and imprecise system that has grown increasingly problematical as the internet has expanded.

The people who were designing the internet – inasmuch as there was any overarching design at all – were rapidly throwing together an experimental network, Day says, not something that anyone thought would still exist more or less in its original format decades later. “Was a lot of thought going into how addressing would work? No.”

Nonetheless, Day says that it didn’t take long for the net’s pioneers to realise there was a serious problem – one that should have been addressed (literally) at the time in 1972, but was not.

At that time, one of the US air force bases using what had begun as a defense department network wanted to build “redundancy” into its location on the network – the equivalent of having two lines running into a single telephone, in case one line went down. But this couldn’t be done as each “wire” had to be given a separate IP address.

“The fundamental problem is that the network doesn’t know that two paths go to the same place. It’s a problem of delivery, not sending,” says Day. “Half of the addressing architecture is missing.”

Numerous solutions have been proposed to patch up the problem, with the most current being a drive to implement what is known as IPv6, a new and expanded form of IP addressing. But Day says IPv6 cannot correct any of the fundamental problems that must eventually be tackled.

The answer to the problem, he says, was actually given in 1992, an alternative way of structuring the network according to open standards he helped to develop for the International Organization of Standards (ISO). But it was rejected by the net’s decision-making bodies.

Since then, he has continued to work on an alternative network structure called RINA (Recursive InterNetwork Architecture), “a complete architecture based on fundamentals”.

The architecture has a single layer, and a single data transfer protocol. It doesn’t need any of the new versions of internet addresses, and, Days says, “It’s inherently more secure.” On RINA, private networks are the norm.

Day says there would be no need for firewalls. Congestion control is also more effective. “The best part is the transition [to the new network] is seamless, and can be done a bit at a time. The public net – the internet – is a layer that would just sit on top. There is no transition, only adoption.”

He seems confident that RINA’s time will come. Eventually, he says, people will realise the existing system is close to collapse. In the meantime, in association with other research teams including TSSG, he is designing and testing RINA.

In his spare time, he maintains a keen interest in history. A personal interest has been cartography, and he has become a leading expert on a single map drawn in 1602, a passion that began with an attempt to research the map’s illustrations of whales.

And of course, he has a special passion for the history of the internet, which he feels has been poorly served by existing books that focus more on individuals rather than the messier and arguably, far more interesting politics and societal context.

“You want to get the ‘who was fighting with who’ down. I think the social interaction, how it played out and affects the outcome, is far more important.”

Some of his own material is already available to researchers and historians – 17 boxes of notes and internet developmental work donated to the Babbage Institute.

“I also do history,” he grins. “I knew this stuff was going to be of interest and I saved everything I could.”

He recalls the excitement and energy of the time, when young researchers of the baby boomer, rock’n’roll generation emerging out of the volatile 1960s, were put in charge of developing the fledgling internet.

“We knew it would be a world-changing thing. We were pretty much on our own and the senior member of our group was just 27 or 28. There were no networking courses – we were figuring this out on our own and we wrote some pretty amazing stuff,” he says.

“We were pushing the envelope.”

For more on RINA: see http://csr.bu.edu/rina/