The idea that voice/data convergence could come from the voice side is not all that strange, but what's unusual is that it comes from someone based in the world of data networking. Mark Slater, technical manager at Extreme Networks, has a valid point though - that synergy and cooperation are better than competition and conquest.

"Why assume that voice on data means the data team takes over? The voice guys hold the key to the true future of convergence," he says. "What frustrates me is Cisco is defining this business by data rules and that's wrong - they don't apply,"

Slater has a number of messages for voice specialists and converging companies alike, and one of them is obviously the need to work together. As he points out, both voice and data networking have very solid technical histories and foundations that aren't going to go away.

After all, we expect the same things from a phone system, whether it is built on traditional circuit switching or IP. That means features such as guaranteed end-to-end delivery, quality of service, availability greater than five nines, the ability to schedule maintenance with little or no user impact, and a lifecycle of 15 years - and if anything, voice networking experts have a better understanding of how to deliver that.

"People need to demand longer lasting prods," Slater adds. "The lifespan of a TDM phone system is 15 years, so IP telephony should be the same. However, when you set TDM in place, typically the application remains static. You tend not to evolve the underlying TDM architecture.

"Most IT kit dates though, because it's built on ASICs. We're now using programmable ASICs so the firmware can be updated, but then the update process has to be non-intrusive because voice needs 24 by 7 availability."

According to Slater, the key for any organisation seeking convergence, initially on voice and data, and perhaps later on video too, is to empower the voice team. He says that voice suppliers such as Extreme's VOIP partner Avaya are finding this a tough one to crack, as the assumption is all too often that the data network is the key element, and they are losing customers as a result to the likes of Cisco.

"Cisco is bamboozling the customer base with science, whether through smoke and mirrors or fact. So the others are uncomfortable selling anything but voice," he says. "I wouldn't say Cisco is misleading people. But because it owns the infrastructure, it can then direct and control the application teams - it's the tail wagging the dog."

And he argues that is the real key to convergence - in the converged network, voice becomes an application, and it's an application with very special demands that the infrastructure has to meet.

"It's common for the applications people to dictate the data infrastructure," he says. "The voice people need to sit with the apps people, not alongside the data people, and they need to be saying 'These are the parameters we expect for our infrastructure application.'

"Voice teams have got a foundation - they've built up a benchmark for stable reliable systems over 130 years. We say, don't change that. When the data guys come along, let them do their job, but do yours too. Drill down into their arguments - how do they guarantee resilience, five nines, etc?

"These guys will only lose their jobs if they allow themselves to. There's no new magic - it's still voice, and the customer's expectation hasn't changed. From the business point of view the expectation should be no different, either.

"At the moment, not only do the data people shift the goal posts, they change the game too. For example, five nines is given for VOIP, but only for the IP telephony gateway. So we say to voice customers, demand five nines availability for your end users, wherever they are.

"Voice teams should be questioning anything that happens in the data network, even if convergence is in the future, for example they should consider the attributes of a data network refresh. Plan for new infrastructure too - ensure that calls are maintained even while the network has a problem. Ask that question!"

Slater adds that there is still a lot that data networking can learn from the voice world. For example, carriers use technology such as SDH or SONET to protect the core by seamlessly re-routing over a ring, so calls are maintained even while the network is converging after a fault. And carriers know how to do capacity planning and ensure they can still carry the load after a failure.

"Genuine capacity planning has never been done to my knowledge by any enterprise I've dealt with," he says. "Genuine means headroom, support for committed rates, and so on. But then, enterprises have never had deterministic applications before."