VOIP reality check
Carriers, ISPs, .coms lead the way, while enterprises lag in voice-over-IP deployments
|
|
|||
|
|
John Chambers uses an IP telephone. So do the rest of Cisco's top executives. In fact, the company has deployed more than 2,000 IP phones in strategic locations like its New York sales office and executive briefing centers used by blue-chip customers.
But the fact that the people who put IP on the map can manage to run toll-quality voice over it doesn't mean that mere mortals can, too. A recent study by Faulkner Information Services indicates that only 7% of enterprises have deployed any voice over IP - generally as small, cautiously implemented pilot projects in least-critical environments.
And according to ICM Global Intelligence's 1999 Voice Over IP Report, the classic "hockey stick upramp" of new technology adoption won't really begin until around 2004.
"Enterprises face the biggest voice-over-IP challenges because of the intermix of a lot of different data structures and protocols in their campus and WAN environments, and because there is no one-stop shopping for voice-over-IP solutions," says Eric Larson, director of networking products for Motorola's Multiservice Networks Division in Mansfield, Mass.
Voice over IP may not be ready for prime time in corporate networks just yet, but it's getting there. And because most experts believe that voice over IP will become a widespread reality sooner or later, it may be time to get on the voice-over-IP learning curve with a pilot of your own - particularly if your company is highly distributed, has employees who move among offices a lot or has seasonally variable voice-service demands.
The early adopters
At this point, the main early adopters of voice over IP are carriers, which are being dragged into voice over IP - kicking and screaming in some cases - by newcomers such as Qwest, which have no circuit-switched infrastructure. The traditional carrier networks were set up to deliver services at 12 to 15 cents per minute, but in order to compete, they have to offer services at 5 cents per minute. Thus, most telcos can't make money without resorting to lower-cost voice over IP.
On the enterprise side, some multinational companies with far-flung subsidiaries are using voice over IP to avoid international long-distance charges. However, most early adopters have motivations other than cost savings. Nortel Networks and Lucent report that initial demand for IP PBXs is coming primarily from small, technologically advanced companies, such as Internet start-ups.
Take Minneapolis Internet start-up Flowers and Gifts, which was spun off earlier this year from a 50-year-old family business.
"Independent florists can't make it with the brick-and-mortar model anymore," says Lauren Christiansen, director of consumer member services for the company. "We decided upfront we would be an Internet-only company."
|
Currently, Flowers and Gifts has regular analog phones hanging off the IP PBX, and its employees interact with customers through Web-based text chat rather than voice over IP. There is downloadable software available that can turn customer computers into IP phones, but the company rejected this approach for the moment.
"I'm not seeing anything yet that is easy enough for our customers to use, and we also aren't sure how many of them have computers equipped with the necessary speakers and microphone," Christiansen says.
Internet start-ups are investing in IP telephony because of what it can do - now and in the future. Bigger, more mature enterprises need to be able to justify such an investment through total cost of ownership. But in hard dollar terms, today's IP PBXs may offer no cost advantage when compared to traditional voice costs, which are plummeting.
However, the issue of whether voice over IP in the enterprise can be cost-justified remains unresolved. Pallab Chatterjee, senior vice president and chief information officer at Texas Instruments, is determined to "build the case for voice over IP on hard dollars," adding that eventual soft dollar savings from applications such as unified messaging "are gravy."
Chatterjee has an easier time than most companies counting these costs because the telecom and data communications groups at Texas Instruments report to him. Voice over IP will be harder to implement in companies that keep the groups separate.
Chatterjee now has about 250 people on IP phones. This deployment is being increased to 1,000 by January as part of a pilot to see how scalable current voice over IP technology is.
"We have now done things like plug an IP phone into the corporate network in Singapore and get a dial tone from a PBX in Dallas," Chatterjee reports. "The network recognizes your phone number, so there are no moves, adds or changes to deal with."
In fact, voice over IP extends all remote-management capabilities of data to voice.
"In traditional voice systems, you can fix 40% to 50% of the problems remotely," says Larry Jones, vice president of enterprise integration services for Williams Communications Solutions in Houston. "On the data side, you can do 80%. Voice is still very people intensive, and people are the most expensive element of the network environment."
Still, many experts agree that early-adopter types are generally not focused on bottom-line communication costs. They are more concerned about the increasing pace of technological change and their abilities to accommodate new developments.
"In the end, what will force the change to voice over IP is the need for a flexible infrastructure," Motorola's Larson says. "Applications are a part of it - but only a part."
For example, CIGNA HealthCare sees voice over IP as a way to help cope with seasonal bursts of customer inquiries. Instead of staffing a central call center for these demand peaks, the company wants to be able to integrate groups at other locations in and out of the call center dynamically as the workload goes up and down. IP Exchange, Lucent's new IP PBX, made this possible, and CIGNA HealthCare served as one of its five initial beta testers.
In the initial trial, an IP PBX was installed at CIGNA's Bloomfield, Conn., campus and overflow 800 calls coming in to the Lucent G-3R PBX were switched to it. These calls then went to call center agents via Ethernet. Later in the trial, calls were routed over an IP WAN connection to IP phones used by remote agents in a southern New Jersey facility..
"Our business managers suddenly understand the reason for voice over IP when they see this," says Len Dorrian, vice president of information management and technology for CIGNA. "It isn't about cheaper phone service. We're after business flexibility. People are the most expensive asset we have, and we have to be able to use them more effectively."
At the end of the test, Dorrian pronounced the preproduction Lucent technology "not quite ready for prime time, but further along than we thought. The IP PBX had some system stability problems that three weeks of troubleshooting by Lucent engineers didn't resolve, and the voice quality on the PC-based softphones varied, according to Dorrian.
Still Dorrian is pressing ahead. "We expect to start deploying something in the latter part of 2000."
Reinventing the PBX
The first round of IP PBXs were the creation of 'Net heads, and as such didn't have much credibility with telecom managers. They also weren't built to scale beyond 100 to 200 users.
Lucent's just-released IP Exchange is the first IP PBX to come from a traditional manufacturer of voice-switching equipment. It supports up to 96 lines, but you can link multiple units together and manage them as a single system.
Nortel is responding with its Internet Communications Architecture, which is scheduled for release at the end of the first quarter of 2000. Nortel's first IP PBX will support up to 2,000 lines.
While no one is claiming the initial products can replace a 50,000-line Meridian switch, how high new IP PBXs can scale in multiple-unit configurations remains to be seen. However, they should ultimately be quite scalable.
"While a PBX is one physical box associated with a certain site, an IP PBX is not," says Christian Renaud, product marketing manager for Cisco's enterprise voice business unit in San Jose. "Instead, it's the Internet model - a lot of distributed devices acting as one system."
This distributed architecture conveys a number of advantages, including fault tolerance. When a corporate campus with 10,000 phones is served by five 2,000-line IP PBXs and one of them goes down, the remaining four can share the load. IP PBXs can also be distributed across the wide area, thus greatly simplifying the provisioning of telephony services throughout a company that has many small branches. The alternative would be to buy 30 separate key systems, then manage and support them individually.
Start-up Tundo Corp. in Boston is emphasizing this horizontal and distributed architecture by calling its first product a "distributed open telephony operating system" instead of an IP PBX. The initial iteration, being beta-tested in four call center environments, runs behind a Definity PBX and extends its capabilities to a second call center some distance away.
For example, Apac Customer Services, a teleservices outsourcing company in Deerfield, Ill., now has 300 agents working out of a second location 100 miles from its main call center and will increase that number to 1,500 over the next year.
"If you were to put in a second call center with its own PBX, it would cost you about $10,000 per seat - $5,000 for the computer and $5,000 for the telephony," says Gordon Payne, executive vice president of Tundo. "We provide the telephony portion by adding ports at $300 to $400 each to the existing Definity PBX, plus $600 per user in Tundo equipment at the host and remote sites. So that's about $1,000 per user vs. as much as $5,000."
Payne says call center managers make particularly good targets for voice-over-IP vendors because they are users who need something and who have a budget.
Installed-base inertia
The Tundo offering, which emerged in September, will eventually evolve into a stand-alone product that can provide full PBX functionality by itself. Meanwhile, it is designed to take advantage of Lucent's installed base of 100,000 or so IP-upgradable Definity switches.
Lucent and Nortel have their own upgrade products, and they have made it clear that their new IP PBX products notwithstanding, this upgrade approach is the path they expect and prefer their customers to take.
In fact, convergence enthusiasts say one of the biggest inhibitors to voice over IP is the billions of dollars carriers, vendors and enterprises have invested in circuit switching. Simply put, those investments are nowhere near recovered yet.
"Nortel and its channel make more money on the upkeep of existing PBX customers than they do on selling new switches, so they hardly want to walk away from the old infrastructure" says Larry Jones, vice president of enterprise integration services for Williams Communications Solutions in Houston. "That's why most of the people who are pushing convergence are on the IP side."
But the new carriers with the pure IP backbones can only push it so far. The local loops are still based on time-division multiplexing.
However, the pieces are coming together to make voice over IP a more viable option. For example, Time Warner Telecom in Denver is building local fiber networks that can provide businesses with an IP connection over the last mile, according to Tony Thakur, vice president of engineering, technologies and planning.
"That's what's been missing," he says. "You never used to hear of local customers needing OC-12s, and now they're asking for OC-48s."
Progress is also being made on the IP phone front, although the devices are still very expensive and don't interoperate across different IP PBX products. Prices should start dropping rapidly now that semiconductor companies have demonstrated highly integrated chipsets for IP phones. A year ago, phone manufacturers had to piece the different components together, which was very expensive.
Also, products are now available from third parties such as PowerDsine for sending enough electrical power over LAN cabling to keep IP phones running during power failures. This eliminates a major enterprise objection to voice over IP.
However, despite all the mergers and acquisitions and talk of technologies coming together, not even Cisco can give you all the end-to-end call management and other capabilities you need to replace your current voice infrastructure with voice over IP.
"I am a little frustrated with the hardware out there," says Sean O'Toole, operations director for Icarian, a Silicon Valley start-up. Icarian is using Interactive Intelligence's IP telephony platform to track phone transactions with customers and provide escalation procedures, but plans to use voice over IP for interoffice trunking and telecommuting applications are still on the drawing board.
"What I really need is the equivalent of a channel bank that lets me plug in some analog phones at the remote office and tie them back into the system at headquarters," O'Toole says. "I need a way to get from traditional analog equipment - switches or telephones - to IP and back at a reasonable density. There are a lot of products that give me 24 analog ports on both ends, but using up a whole T-1 for 24 analog lines is a real waste. I should be able to get 50 to 100 lines and still have room for a data connection. It's such an obvious product that I don't know why it doesn't exist."
Breidenbach is a freelance writer based in California. She can be reached at sbreide@aol.com.
RELATED LINKS
Voice over IP implementation in the enterprise is lagging behind expectations partly because several factors - deregulation, economies of scope and scale, and the outsourcing trend - have shifted the momentum to the service provider market. Network World, 11/29/99.
QoS Questions
Out on the WAN, guaranteeing sufficient quality of service (QoS) for voice applications is still a problem in the public Internet, where service providers can't control the infrastructure from end to end. Network World, 11/29/99.
Don't look to the enterprise to drive voice over IP
If voice over IP doesn't provide a clear cost benefit, then it has to provide an application that gives the enterprise a business advantage. Network World, 11/01/99.
Voice over packet: But which packet?
Despite the hoopla surrounding voice over IP and its purported savings potential, the companies currently running packetized voice tend to be using frame relay instead. Network World, 07/21/99.
Voice over IP coming in loud and unclear
Columnist Dave Kearns believes voice over IP doesn't make sound business sense. Network World, 06/07/99.
Voice over IP Netresources
Primers and more. Network World Fusion.
Carrier-Class Voice over IP
A whitepaper from Compaq's Digital division.
Feedback
Tell us your thoughts on this article or the issues it raises.

