O'Dell: Voice and Data Don't Mix
"Data and voice simply don't mix," said Michael O'Dell, president of Compass Rose Labs, a telecom consultancy. "You simply can't do both."
While there are networks that can handle data and voice traffic, a company that sells and supports data and voice services equally well is "a myth," according to O'Dell, the former UUNet chief scientist that spoke at Opticon 2002 on Tuesday.
O'Dell's beef is that the services are so different that they need to be offered by different companies. Having a traditional phone company try to set pricing and deliver such a product as IP data just causes problems.
The structure of a company is reflected in the structure of its network, he says. In the incumbent networks, a large number of people are required to keep it running. Internet and data networks, however, were designed to avoid human intervention, he says. "Having to physically touch something in a data network is looked at as a bug, not a feature."
Another difference: Data networks have a distributed control system with each router telling the next one what to do. Phone networks, on the other hand, are centrally switched and controlled and the connections are point-to-point.
"The Internet represents the death of centralized planning," he says. It evolves biologically, meaning that most experiments in new features won't work, but they need to be tried in mass and in parallel, he says. This differs from the lock-step way that phone networks roll out new services via a centralized "Department of New Ideas," O'Dell says.
"Nature works by doing millions of experiments in parallel and most of them die," he says. "That's not how centralized planning works."
The way to develop new services for communications networks is to find a way to let millions of people experiment to see what works, according to O'Dell. In the computing world, this happened when the industry went from having dozens of people maintain a centralized mainframe computer to having millions of people try things with their own personal computers.
Contributing to the industry's downfall, O'Dell says, is the fact that the companies so hell-bent on the success of converged voice and data are trying to solve an economic problem (how to cut the costs of operating a network and providing services) with technology (faster and cheaper hardware).
In voice networks, hardware costs are a fraction of the total network's cost, he says, yet too many companies have tried to make phone networks more profitable by offering faster, cheaper hardware. What should be happening is an attack on operating costs, he says.
Converged companies selling services on converged networks also have a problem, in that the economics of offering voice services and data services are completely different. "For an IP product, the cost of goods is a huge percent of the price," O'Dell says. "For voice products, all you really need to do is cover the cost of the ad campaign."
Of course, O'Dell's analysis of the way phone companies work -- and his comparisons of phone companies to Communist Russia -- are not entirely surprising (see Can Customers Take Back the Network?). His previous company, UUNet, was a data-only company that eventually became part of WorldCom Inc. (OTC: WCOEQ), a traditional long-distance phone company.
So how do we recover from all of this alleged wrongheadedness? Don't ask O'Dell. Like any good speaker, he left the audience here with more questions than answers.
"If you're looking for the silver bullet to turn this industry around, you won't find it here," he says. "If I was that smart, I wouldn't tell you."
— Phil Harvey, Senior Editor, Light Reading