VoIP

Consider VoATM, Not VoIP, for Small Businesses

The voice over ATM (VoATM) market is not for everyone. But if you're an ISP or CLEC and already have backbone or NAP contracts, you just might find (if the stars are aligned) that you can sell small business VoATM at a profit.

by Max Smetannikov [October 28, 2002]

It's been some time since the Internet telephony hype first hit the wires. At the time, some ISPs wondered whether one day they might need to sell voice services to retain there customers. Fast forward to today, and ISPs—especially those catering to business customers—are losing premium customers to voice-enabled competition.

So now that the business climate has changed 100 percent and ISPs trying new things are frowned upon, the time has come to go into the voice business. ISPs that once brushed off telephony as an inconsequential or even impossible service now see competition from service providers who know how to sell voice and data over one pipe.

At least that is the situation that Richardson, Texas-based Inturnet has encountered on its home turf. After seven years in business as an ISP catering to the enterprise crowd, and after making a painful transition from reselling DSL to reselling T-1 connectivity, Inturnet started losing customers to competitors selling voice and data bundles.

"We were talking for a long time about doing it but now the time seems right," said Tod Weber, Inturnet president, commenting on Intrunet's latest foray into voice over packet.

Five years ago, an ISP like Inturnet would have looked closely at a classic voice over IP (VoIP) solution, complete with a line of equipment that would packetize voice at the customer end. The ISP could also have looked to interconnect with a VoIP wholesaler like ITXC or DeltaThree, packetizing voice services at the ISP end. At a minimum, it would offer its customers a customized version of click to talk software available from companies like Net2Phone and PC2Phone. The ISP would also have looked at a voice over DSL solution if DSL resale was available, using hardware from CopperCom or JetStream (now Paradyne) at the customer end.

While all of the above solutions do have a niche in the ISP business, in the meat and potatoes land of $500 a month T-1s, voice has to come on top of a service that already makes money.

The economics in Richardson are such that T-1s can be had wholesale from a CLEC, always good news for ISPs that almost never get the right product at the right price from incumbent phone companies. Intrunet made a business out of retailing T-1 pipes it buys wholesale from El Paso Global Networks for $175 a month (the company is a subsidary of El Paso Corporation, an energy group. Even loaded with backbone bandwidth and overhead costs, this entry price still leaves enough margin to run a profitable business.

At least it did before customers started to churn away to ISPs that offered voice. Since Intrunet wasn't about to reinvent the wheel in order to add voice to its services lineup, it had to figure out a way to do voice over T-1s. Given that T-1s were invented by phone companies in the first place, that proved to be easy. However, Intrunet had to nix VoIP and use VoATM.

"We are ATM all the way," says Weber.

The bottom line reason Intrunet chose to abandon its IP roots is simple. An IP-encoded voice call takes up 100 kilobits, where as an ATM-encoded call takes up 40 kilobits. This probably doesn't matter for customers running fiber pipes into their offices, but for smaller firms using a T-1s and looking to save on the $40 a month they are paying individually for five phones or more—this is a big deal.

Like Internet lines, phone lines are routinely oversubscribed. At 100 kilobits per call, an average volume of calls can easily kill all available bandwidth on a T-1. "This is not an option for many businesses, since e-mail is no longer a convenience—it is a necessity," Weber said.

To make the VoATM set up work, Intrunet bartered with one of its customers for some termination equipment, getting its hands on a CopperComm ATM to GR 303 converter and an Alcatel switch that moves traffic from individual T-1s onto voice and data networks. At retail prices, this equipment alone would cost $50,000, Weber estimated.

To terminate voice traffic, Intrunet signed up with Timer Warner Telecom. It already had an IP backbone connection through InterNAP.

The only sticking point with the service was the customer premise equipment. Intrunet uses Polycom aggregators that support from eight to 24 telephone lines. However, those boxes start at $900 in their cheapest configuration. Asking customers to pay that upfront would make the sale of the new service difficult, but giving the equipment away as some competitors do with similar Cisco equipment would make these customers pay for themselves in three years instead of the three months Intrunet requires. The company is still researching strategy, but offering customers a lease of some sort sounds like the right approach, Weber said.

Of course, voice over packet services are not for everyone, and require a major commitment from the service provider. This is not the kind of service that can be offered as an afterthought.

Eddie Dreger, a networking consultant with Brotsman & Dreger's EverQuick Internet Division, recently blasted an ISP executive on an ISP-Lists discussion list who asked whether voice is a good business to develop as a way to avoid wasting extra bandwidth.

"I responded to a poster who had a T-1 and wanted to do voice to cash in on the excess bandwidth. However, the costs of offering general long distance service would exceed that of the wasted T-1 bandwidth," Dreger explained. Indeed, if you consider Intrunet's experience, the stars have to be aligned just right in the local market for voice over anything to be a goods business model.

Also, the odds that large service providers will offer meaningful competition with integrated voice and data products in non-NFL cities are declining fast. Sure, a range of data CLECs like Primus and Focal Communications sell such services, but they typically are not interested in customers that have five phones. So this market is wide open for ISPs—if they can make any money out of it.

—End