The technology is working beautifully at the University of Guelph, but not every installation will turn out as well as this one did.
September 1, 2004
When the University of Guelph decided in early 2001 that its 15-year-old phone system would not last much longer, a Voice Over Internet Protocol (VoIP) system seemed like the logical successor.
For a couple of years VoIP had been gaining attention, promising cost savings and convenience, and the technology looked to be mature enough to do the job.
“We needed to look forward and choose the technology that was going to be mainstream in the longer term,” says Kent Percival, the university’s manager of computing and communications services development.
As a three-year project to replace virtually the entire campus phone system nears completion this fall, Percival is more than satisfied with the choice.
The university spent about $16 million on equipment and network upgrades, he says, and for the price, gained an improved data network along with its new phone system. When you allocate some of the cost of network upgrades to data, Percival maintains, the VoIP system comes out costing no more than a traditional PBX, while offering more flexibility for the future.
However, not every VoIP implementation turns out as well as this one did and while the organizations that are not happy are usually reluctant to talk to the press about where their projects went wrong, what made Guelph’s implementation work says a lot about why VoIP isn’t for everyone.
An old data network
First, Guelph realized from the onset that its existing data network was not up to VoIP. Most of it was as old as the phone system — Category 3 cabling installed in the mid-1980s. The campus would have to be rewired to support VoIP.
That would have been a deal-breaker in many cases, but at Guelph it made sense. The university needed a better data network anyway, and because the old phone system could last another three years or so, there was time to do much of the re-cabling as part of renovation projects aimed at preparing for the “double cohort” of incoming students caused by Ontario’s elimination of Grade 13.
There were also concerns about the reliability of voice over IP. “A number of executives were hearing that VoIP was not going to deliver five nines reliability,” Percival says. Those concerns were legitimate. He points out, though, that while the old phone system was already having reliability problems, it would sometimes take as long as one to two hours before complaints began to flood in.
Meanwhile when the university’s e-mail system failed or was down, the calls would start pouring in within 20 minutes.
Given such strong reliance on e-mail, and the use of e-learning in classrooms and laboratories, “there are a whole lot of reasons why we need a data network that is approaching the reliability of the phone system,” Percival says.
So it made sense to spend on an upgraded data network that would also support voice with acceptable — though maybe not five nines — reliability.
This type of scenario does not fit everyone, which is why even some vendors of VoIP equipment admit it’s not the right answer in every situation.
That is no surprise to Elizabeth Ussher, vice-president at Stamford, Conn.-based Meta Group Inc., who says the research firm predicted a VoIP backlash a few years ago.
Vendors coming from a data background “didn’t understand the complexity of voice,” Ussher says. “It was kind of a big surprise to them that oh, people really do use more than five features, gee I think I really do need a message-waiting light.”
For the most part, the features of IP phones are now comparable to those of traditional systems, but for many organizations the return on investment in VoIP is still hard to find.
Ussher says — and many in the industry agree — “there is little reason for any organization to move to VoIP unless it has to do something anyway because its PBX is on its last legs or its Centrex contract is running out. “If things are pretty much status quo in my enterprise, I have no reason to go through this effort.”
Making the right choice
Even for an organization that has to replace its phone system or is starting from scratch, VoIP may or may not be the right choice — or may be right for parts of the organization.
“We’re incredibly excited about what IP telephony does,” says Tracy Fleming, convergence specialist at Avaya Inc. in Toronto. In some cases it may make sense to use VoIP throughout a building or campus, Fleming says, in other cases it makes sense only in pockets, particularly in older facilities.
Phil Edholm, chief technologist and vice president of network architecture at Nortel Networks Corp., says VoIP is a good fit for organizations with a lot of telecommuters, for providing more economical voice and data service to branch offices, in environments where phones are frequently moved around, but can be hard to justify when infrastructure costs are high and the benefits harder to realize.
Of course, Avaya and Nortel are both companies with a background in traditional phone systems. Data-oriented companies are far more enthusiastic about VoIP. Brantz Myers, national manager of enterprise marketing for Cisco Systems Canada Inc., says “this will be an IP-only market in the next couple of years.”
Edholm does not rule out IP prevailing eventually, but it certainly is not going to happen within the next two years.
He says organizations will switch, not because they want VoIP rather than traditional PBX technology, but because they want capabilities the new technology offers just as they switched to IP for data in the 1990s because of the World Wide Web. “What are the applications?” Edholm asks. “I don’t have a clue.”
Until those applications emerge, voice over IP is not the only option, though it can be a good fit in the right circumstances and work well if implemented properly.
Mobile workers are among the best candidates for voice over IP because it allows them to take their phones with them wherever they can get a broadband Internet connection. Edholm, for instance, says his calling card bills ran $250 to $300 per month until he started using an IP phone on the road about 18 months ago.
He now plugs his IP phone into his home broadband connection or a high-speed connection in a hotel room and it looks like another extension on Nortel’s office phone system. His calling card bills — still covering situations when broadband connections are not available — are down to $20 to $30 per month.
Floyd Marinescu, the Toronto-based creator and general manager of media properties for The Middleware Co., a Mountain View, Calif.-based knowledge network for software developers, is an IP proponent.
“I’m a consultant primarily with U.S. customers and I wanted to cut down my long-distance bills,” he says.
Marinescu recently signed up for a VoIP service from Primus Telecommunications Canada Inc. that allows him to make and receive phone calls over a broadband Internet connection.
Services such as Primus’ Talk-Broadband, which is offered by carriers on their own networks, are proliferating.
Bell Canada has a Managed IP Service for businesses and is running trials of a consumer VoIP service with plans to address the small-business market. Lawson Hunter, executive vice president of Bell parent BCE Inc., says Canadian telcos and cable carriers will pile into this market over the coming year.
Matt Stein, vice president of new technology and services at Primus, says the benefits include reduced cost and the ability to have your phone calls routed to wherever you are.
For Marinescu, the specific impetus for getting the Primus service was a business trip to Spain. “People can call me in Toronto and I can pick up over there,” Marinescu explains. “It was like having a phone with an extension cord across the Atlantic.” Marinescu estimates he saved $1,000 on long-distance charges during the three-week trip.
Five nines defined
There were some initial glitches getting his IP phone to work over a slow DSL connection with a firewall. “It certainly wasn’t flawless,” Marinescu admits, “but it saved a lot of money and it worked well most of the time.
That statement sums up the most promising cases for VoIP. When there are significant cost savings or other benefits to be had, users will accept minor tradeoffs in reliability.
And reliability is an issue. Edholm says traditional phone systems deliver five nines reliability meaning they work 99.999 per cent of the time.
Most data networks can provide three to four nines, or 99.9- to 99.99-per-cent reliability. It sounds like a small difference, but in terms of how much the phones are not working, Edholm points out that “the difference between five nines and three nines is the difference between five minutes a year and eight hours a year.”
However, both Edholm and Fleming say data networks can be made to provide five nines reliability or very close to it.
The issue is what it takes to do it and whether, once you consider the cost of doing that, VoIP still looks like an attractive proposition. Edholm offers a rule of thumb: for every dollar you spend on VoIP gear, expect to spend two to five dollars on upgrading the data network to support voice.
As a by-product of that you may get improved data performance, Ussher says. But “if in fact you don’t have the capacity and ability to implement a secure reliable voice-quality data network, then you really should leave things on two separate networks.”
Ussher says failing to prepare the network properly is the number-one cause of VoIP problems. The number-two cause, she says, is people problems. If your staff are not prepared to implement and run a VoIP network, there will be problems — unless you outsource the job, in which case “it breaks your return on investment model badly.”
On the other hand, she adds, one of the great benefits of converging network infrastructures may be the associated convergence of staff into one group instead of separate voice and data teams.
A key factor in making VoIP pay is moves, adds and changes, Ussher says. An IP system makes it easier to rearrange phones as employees move around. This is common in some organizations and less common in others and in many places it happens with some phones and not others.
Often some sort of hybrid of IP and traditional phone technology is the right answer. The Ontario Rental Housing Tribunal is an example.
Established in 1998, the tribunal quickly discovered it had underestimated the call volumes it would be handling and its original phone system was not up to the task, says Brian Opitz, manager of corporate services. The overloaded system was dropping calls and the reports it produced were confusing and of little help in managing the system.
The tribunal installed an Avaya Definity Enterprise Communications Server to route calls to agents across its 14 call centres (now reduced to eight). But while the Definity system uses IP connections to send call information to agents’ PC screens and provide supervisors with online monitoring data, the calls themselves still go over regular phone lines.
At least in 2002, Opitz says, voice over IP hadn’t developed to the point where it was really reliable, and the tribunal had an existing phone infrastructure that could do the job. “If we could use the hybrid, we would get the best of both worlds,” he says.
More organizations are coming to this realization, Fleming says. Whereas a year or two ago more customers wanted pure VoIP or none at all, “in the last six months, eight months, we’re seeing much more open-minded approaches to this technology.”
Ussher says the number of IP phones sold has been growing, and the number of failed VoIP implementations has been declining. The feature sets of IP systems are now comparable to traditional phone systems, she says, and organizations are learning something about how to make VoIP projects work.
“The first time you try to make a souffl you don’t do very well,” she observes. “The next time you do a little better.”
Grant Buckler is a Kingston, Ont. freelance writer and editor who specializes in IT and telecommunications issues. He can be reached at firstname.lastname@example.org.