When CCL Industries, the Canadian retail goods packaging company, made the decision last year to move from traditional circuit-switched telephone systems to IP telephony, CIO Akhil Bhandari insisted that caution was the better part of valour. He would only commit to making the change at the company’s relatively small Toronto headquarters. It was a good decision. CCL did not have a happy initial experience with VoIP.
The company was hit by two or three outages in the first weeks after the cut-over in January this year. In one case, his staff was able to get the system back up and running just in the nick of time for a crucial teleconference. There were cost overruns beyond the initial fixed price the vendors quoted because the data network needed to be upgraded. And some key features — including caller ID, one the company was particularly looking forward to having — did not work at first.
“Because of our experience during the pilot, we are not rushing out to replace phone systems in all our other facilities,” Bhandari says. “As long as the old system works, we’ll stay with it. If it drops dead, then we’ll make a change.”
The system vendor and integrator — he doesn’t want to name them, but they are major market players — eventually stabilized the system and got all the features working. The impacts for CCL were ultimately not huge: irritation, some lost productivity, embarrassment. It could have been worse. Two years ago, Merrill-Lynch, had so many problems it scrapped major IP telephony installations from a leading vendor in New Jersey and Tokyo and replaced them with a mix of IP and traditional telephony equipment from a different source.
Not that migrating to IP voice is always or even often a bad idea. But it’s a difficult transition to make. “It’s so inherently difficult it makes you wonder who ever came up with the crazy idea of trying to run voice over a data network in the first place,” says John Riddell, a senior analyst with Decima Inc. “Data networks were not designed to carry voice. They don’t have the built-in qualities to do it. So it’s hard, and despite all the help you can get from vendors, it remains hard.”
It can be done, though. And is being done — successfully — by more and more companies. According to the Enterprise Communications Association’s (ECA) 2005 ECA Converged Communications Systems Market Report, IP PBX stations accounted for about 48 percent of shipments in the US in 2004, TDM stations about 52 percent. The report’s author, Allan Sulkin, president of TEQConsult Group, predicts that IP PBX station shipments will be at 88 percent of total by 2012.
In its 2005 IT Priorities report, London, Ontario-based Info-Tech Research Group reports that 23 percent of the mid-size enterprises it surveyed already have some form of IP telephony in place, 16 percent are planning to implement VoIP in 2005 and an additional 24 percent said they want it in place within three years.
There are good reasons they’re doing it: cost savings, including on long distance charges, by running inter-office calls over an existing IP wide area network, improvements in business processes — most notably in call centres — plus difficult-to-quantify office worker productivity increases delivered by new IP-based features such as presence management, unified messaging and find-me-follow-me call routing.
According to a study by Info-Tech research analyst George Goodall, switching from traditional Centrex to either an IP Centrex service or premises-based IP PBX equipment in an office with 24 stations results in cost savings of as much as 40 percent over five years. Making the switch from TDM premises-based systems to IP systems would save “at least 20 percent.”
“The real sweet spot for VoIP,” Goodall says, “is companies that are expanding quickly — either their head count or just the number of handsets they need to support, or they’re increasing the call or data volume their telecom systems need to support. That’s where there’s a strong play for VoIP.”
Notwithstanding the CCL experience, the risks involved in making the transition to IP have diminished as vendors, integrators and consultants learn how to lead customers around the pitfalls. We convened an informal panel to point out some of the danger areas. Our panel: Riddell, Goodall, Wayne Ingram, a Toronto-based partner with Accenture, and Jerry Sparling, vice president of customer quality, services and support at Mitel Networks Corp. They identified at least seven key areas that companies need to focus on.
Business goals
Understanding your business goals is the first and most important because it determines decisions you’ll make later about network architecture, vendor selection and other issues. Are you only migrating to IP because your old PBXs are dying? If that’s the case, one crucial part of the planning process is to itemize the features of your existing telephone system that the new system must replicate. As Riddell notes, not all VoIP systems will be able to duplicate all traditional PBX features.
Are you hoping VoIP will reduce costs through toll bypass, savings on moves, adds and changes and possible IT department head count reductions? Don’t take the last two for granted, Riddell warns. They do not come automatically. Changes in internal processes and corporate culture must come first. And if you expect savings from toll bypass, you will have to ensure that not only the LAN is up to speed, but also your WAN.
Are you hoping to leverage new features — presence management, unified messaging, integrated voice and Web conferencing, IP-based video conferencing? They may deliver real benefits, but again, not automatically. Extracting promised benefits requires careful planning and management. And applications like IP video will have significant impacts on network capacity requirements.
“You need to know what path you’re on from the beginning,” says Accenture’s Ingram, “because that drives the rest of what you do.”
Team readiness
There are two potential IT department problems: skills — or absence thereof — and turf contention. As for the first, Riddell says, “You would have thought we would be through that by now, but in talking to customers, it’s clear this continues to be an urgent problem.”
It can be solved by implementing training programs, of course, but as Mitel’s Sparling notes, “Most people working with IP voice for the first time underestimate the skills gap.” Team members need to be experienced with esoteric networking functions such as voice and data grooming, packet priortization and configuring routers for voice.
Training costs must be factored into the business case, Goodall stresses, and they can rival technology costs. “Knowing what you have in terms of human resources is crucial,” he says.
The problem is often aggravated by turf conflicts. Data people know little about voice, telecom folks little about data. VoIP requires both skill sets, plus special skills besides. It’s clear the two groups must be merged, but don’t underestimate the problems involved. Riddell has seen companies make compromised vendor selections because battling voice and data groups couldn’t agree. “This remains a prominent difficulty that needs careful attention,” he says.
If you’re using a systems integrator, don’t assume these problems all go away. Riddell has seen cases where a senior technical person with impressive skills is brought in during the selling phase, but is nowhere to be found during implementation. “Know exactly who from the integrator is going to be involved,” Riddell says. “Talk to references.”
Technology architecture
If your company has multiple sites, do you want one big switch at head office and gateways at satellite offices, or smaller autonomous swit
