Networking vendors and telecomms carriers are all putting their spin on it. Industry analysts and members of the trade press are all talking about it. End users are all wondering about it. The 'it' in question is voice/data convergence -- the buzz phrase for 2000 and the next few years to come. Linda Stuart reportsThe idea of combining voice and data on a single infrastructure is not new - ISDN, frame relay and Asynchronous Transfer Mode were all designed with convergence in mind. But, the IP (Internet Protocol) revolution has the networking industry re-examining its convergence strategies, leading many to conclude that IP - with its ubiquity and standards-based approach to communications - will be the underlying protocol for converged networks of the future.
Most of the research and development dollars that vendors spend go to IP-related projects. Traditional data networking companies working on new features for IP equipment and software include Cisco Systems, 3Com and Newbridge Networks.
While Lucent Technologies and Nortel Networks/Bay Networks continue to develop their traditional circuit-switched equipment for the carrier market, both companies have made significant IP investments as well.
However, although these companies - and Cisco especially - are urging customers to architect their networks around IP only, analysts warn users not to get caught up in the "it's the IP way or no way" religious war.
Michael Speyer, program manager for research company, The Yankee Group, said: "The long and short of it is that most of the service providers and carriers we speak to are putting ATM in the core of their networks today" "So certainly, while you might have IP at the edge of the network, today they're going to use ATM in the core, so it's a bit of a mixed bag," Speyer added.
Bob Hafner, vice president and research director at Gartner Group, agreed.
"There are going to be carriers that will offer voice over IP over ATM. Underlying that IP infrastructure will be an ATM infrastructure," Hafner said.
"We're right at the verge of the rollout of these services where the customers are going to mix voice traffic with data traffic and it's going to run over a single infrastructure. And right now, the preferred or better-controlled infrastructure is ATM. The control mechanisms, the ability for quality of service, some of the compression methods are just a little bit more manageable on ATM.
"I think that IP will catch up, it's just a matter of whether it's one year, or a year and a half, where the capabilities will be so close that I suspect it won't make a whole lot of difference any more what the underlying infrastructure is."
In the US, long-distance providers AT&T and Sprint have announced plans to deliver broadband voice, data and Internet access on one line, but AT&T's Integrated Network Connection and Sprint's Integrated On-Demand Network services are still in the trial stage.
Some cable companies are experimenting with IP telephony. But they, along with other Internet service providers are still a few months away from rolling out voice-over-IP services.
Some of the difficulties currently facing multiservice carriers in general, such as call distributing and billing, are compounded when applied to packet-based networks.
"Right now, when you interconnect the circuit-switched network between two different carriers, the way you [distribute] traffic with CCS7 (Common Channel Signaling 7) is very well understood," Hafner explained.
"But how do you hand off these things, and do all the billing for all these services, and track when the call begins and when it ends because it's not like circuit-switched (networks) where there's a continuous set-up and you know when the set-up is done. Here it's just packets flowing back and forth.
"These things are being built and being worked on, but they're certainly not trivial and there is a huge amount of work yet to be done," Hafner said.
At some point packet network operators may decide it is cheaper to give customers a flat rate and unlimited service, than having to count all the packets transmitted or minutes of traffic back and forth, because the cost of actually monitoring that may be more expensive than the actual transmissions.
Much like client/server revolutionised enterprise application architectures, voice and data convergence has the potential to completely change the way networking and communications applications are built and delivered.
And, much like client/server was touted to save companies money but in the end turned out to have significant hidden costs, convergence's reputed promise of reduced costs due to simplified management, streamlined infrastructure and reduced long-distance charges may not tell the whole story.
There's nothing to say that paring separate voice and data networks down to one converged infrastructure means simplified management, according to Gartner's Hafner.
"Ask anyone who runs a LAN network, with complex applications and many users on it, if it's easy to manage that compared to the typical PABX system," Hafner said.
"PABX systems are pretty well understood. [PABX administrators] know when there's a problem, they know how to fix the wires, they can pinpoint a problem pretty quickly, they know when there's a bad card in the PABX. There are not that many places where a problem can occur."
But the ease of management associated with a PABX-based voice network goes out the window when voice is added to a data network.
"You're going to need some pretty sophisticated tools to figure out, when someone picks up the phone and they can't get through, why that is, because it's not a physical circuit-switched environment any more," Hafner said.
He said although the capital cost of equipment for doing voice over a data network may be cheaper than traditional voice gear, users may find the tools lacking at the moment for management, problem resolution and ongoing maintenance of a converged network.
According to Hafner, a total-cost-of-ownership analysis that includes not only the cost of equipment, but operations and support as well, may find that running voice over a data network costs the same or is a bit more expensive than continuing to operate two separate networks.
"But on the other hand, we suspect that you'll have no choice, that there will not be other alternatives," Hafner said.
"All the new applications are going to be designed for [a converged network]. It will become a necessity, as opposed to 'I want to do this to save money'."
The good news for data network managers, and consequently the bad news for telecomms managers, is data experts will have the upper hand when it comes to taking on the responsibility for managing the converged infrastructure. They already understand the most complicated half of voice and data convergence.
"Because it is a data network at that point, you're running an application that just happens to be called voice, as opposed to SAP or Baan or PeopleSoft," Hafner said.
Some of the vendors approaching network convergence from the data side have gone so far as to say customers should stop investing in their legacy PABX equipment or throw it out altogether - this was the main message delivered by Cisco CEO John Chambers earlier this year.
"If Cisco has its way, it will be an IP world," Yankee Group's Speyer said.
"I don't think that any vendor can articulate quite as strong a convergence story in the IP space as Cisco can," he added.
However, Gartner's Hafner said users should not be too quick to toss out their PABXs.
"There are a lot of issues to think about. For example, what about the phone you have on your desk? If you throw away your PABX, what happens to that phone?" Hafner asked.
"From [Chambers's] perspective, the first thing he wants to do is get you to replace all your phones with Ethernet phones, instead of phones that you might have that are PABX-specific. Clearly, he wants to roll over the entire infrastructure."
Another company with an Ethernet phone strategy is Lucent, which recently announced trunk cards for its Definity PABX system that support Ethernet connections, allowing calls to be routed to Ethernet phones.
Using a product such as Lucent's trunk card, customers could gradually replace their current phones with new Ethernet phones through attrition. For example, if a customer had 200 phones at one location, it might decide to install Ethernet phones the next time some of the phones needed replacing, perhaps in batches of 20 at a time.
"Realistically, you're not going to forklift everything out of everybody's locations and replace all these things," Hafner said. "Most people, when they build a business case for voice systems, in the past have generally written the business case with the expectation that the life expectancy of the equipment is seven years. Anyone who has written one in the last two years is probably still going to have their PABX five years from now."