DSL

GWI's Big Lucent Buy

GWI, one of the oldest ISPs on the planet, has just invested in a significant DSL infrastructure upgrade. We spoke to the company's founder and president to find out why.

by Alex Goldman ISP-Planet Managing Editor [March 29, 2004]

Fletcher Kittredge, founder and president of Biddeford, Maine-based GWI, says business is booming. Kittredge, and his company, had a head start, and therefore an advantage, in building Internet service. "In 1984, I got out of college and went to work for BBN, which built the ARPANet," he says.

Having helped build the precursor to the Internet, he left BBN in 1993 and started GWI in 1994. But even Kittredge had no idea which applications would be popular. "I suddenly realized when it showed up on The New York Times that they really were going to commercialize the Internet," he laughs. "When I founded this company, I thought the Web was a toy. There's FTP and e-mail and gopher, which are serious, but this glitzy Web stuff probably won't amount to much."

So now that he's upgrading to Lucent's video-capable DSLAMs, he's cautious about trying to predict what consumers will want. He does, however, know what the key questions are. "Will there be an explosion of channels? Will video phone become big? It doesn't seem like it, but we don't know yet how it's going to turn out."

The company's been in broadband since 1996, when it signed agreements with two local cable operators, Metrocast and Susquehanna Communications. It's been supplying DSL since 2001, but it initially deployed SDSL infrastructure and had to switch equipment providers when ADSL became the accepted consumer standard.

Kittredge is happy with Lucent. "Our major problem is that vendors are not building for the U.S. market because the U.S. market is behind, technologically," he explains. "Lucent seems to have gotten on the IP bandwagon. The IP2000 is actually pretty good. It basically takes the Stinger and makes it much more useful."

Kittredge admits that Dave Burstein (principal of DSL Prime, whose columns are reprinted on this site) believes the backplane of the IP2000 will prove insufficient, but notes that it will only be insufficient if everyone is using video at the same time.

This would assume that a) every DSL subscriber also subscribed to video and b) that every DSLAM was full. As Kittredge points out, it's still too early to say how great demand will prove to be.

He's making the investment now, because he believes DSL is better than all other available broadband technologies. "The best place to be right now is on the telephone network, and DSL, worldwide, is blowing the socks off everything else (except maybe emerging hybrid fiber coax because that can do 100 Mbps)."

However, ISPs don't have to pick one technology. Unlike the dinosaurs, ISPs can do everything at once. "Telephone companies need to say DSL is better, and cable companies need to say cable is better, but we don't have to choose."

In fact, that's part of what's so cool about the Internet. "People get caught up in whether you're doing wireless, or DSL, or dialup, but that shouldn't happen. You do what's cheapest this month. The Internet is successful precisely because it's media independent. We do our job. We go and look and find stuff that has the best price point to get what the customer needs."

Right now, that means Lucent. "Right now, we decided to buy Lucent's top of the line stuff because it's ADSL 2 capable. We'll sell it at $29.95, and at that entry point it's going to be 3 Mbps or 4 Mbps within a few months. We don't need to turn on ADSL2+ yet, even though it's reasonable now to do video over DSL."

Instead, Kittredge expects video over DSL to get cheaper as it is more widely deployed. He points out that VoIP was expensive when it required big iron, but got much cheaper when most of the heavy lifting could get done in software. He expects the same to happen with cable headends.

Another obstacle to deploying video is choosing a set top box manufacturer for the CPE. Kittredge notes that several manufacturers have systems already deployed, but complains that certain standards need to be clarified.

Kittredge is particularly enthusiastic about the potential of multicasting, where one IP stream gets sent down any one pipe, but is duplicated at various points where the path forks (for example, a DSLAM with 100 customers would receive 1 copy of stream and broadcast 100 copies).

There are still certain problems that need to be solved before multicasting can work perfectly. It works, he says, if everyone tunes in and stays connected, but it not as good about people flipping channels (adding and dropping from groups). Also, there is a potential IP address problem.

Meanwhile, GWI has to continue to keep costs down and fight regulatory battles. The GWI network, Kittredge says, uses ATM only for the link between the CPE and DSLAM. The upstream is Ethernet, starting with the GigE port on the back of the IP2000.

Making sure the government doesn't prevent him from doing business does, however, remain a headache. "Being a facilities-based DSL provider is not a bad business, but you do have to work the regulatory issues. Yes, that sucks. But life and business are hard. Almost every sector of the global economy is competitive and that's just the way it is."

He is surprised that the phone companies see him as their enemy. "The phone companies make a fair amount of money off the DSL we sell, and our customers remain Bell voice customers. On the other hand, if they leave for cable, they usually become a VoIP customer." He would know.

—End