AT&T'S major outage of dial-up Internet access last week, which pointed up the fragility of some Internet connectivity methods, may have implications not only for remote workers who rely on WorldNet service, but also for corporate adoption of AT&T's virtual private network offerings.
AT&T's WorldNet service on December 1 was crippled - for the better part of a business day - because of faulty routing information provided by another ISP, company officials confirmed.
While the direct impact on corporate users is hard to gauge, the outage highlighted a disconnect between carriers' promotions of such services as suitable for corporate needs, and the services' potential for failure.
This was particularly apparent in the case of AT&T's outage, because it also stymied access to one of two VPN services that AT&T now offers - and it came just as AT&T has begun putting its weight behind VPNs, according to Stephen Harris, an analyst IDC.
"AT&T has added a number of new enhancements to its VPN service, which until recently has been marginal compared to other carriers," Harris said.
Affected by last week's outage were many corporate customers using AT&T's VPN service but not those using a second service - Global VPN - which grew out of AT&T's purchase of IBM's Global Network.
However, AT&T officials downplayed the impact of the outage on corporate customers - characterising the failure as nowhere near as tragic as the company's frame-relay outage early last year.
"This is a very different type of thing, since it involved mostly individual users and not our MIS core customers," said Rose Klimovich, director of AT&T's Global IP Network Services.
AT&T initially identified and reported the outage as one related to problems with domain-name server emulation.
"We made multiple attempts to diagnose a DNS problem," Klimovich said. "When that wasn't working, we decided to take a bigger look at what was going wrong."
Klimovich said the company later determined that one of its ISP peering partners had inaccurately entered addressing information. AT&T would not reveal the name of the ISP.
"What happened is that another ISP advertised part of the AT&T address block as if it were theirs. There could be a lot of explanations for why that happened, including someone mistyping that information," Klimovich said.
AT&T restored service at about 4:30pm Eastern Standard Time last Wednesday, after many of its dial-up users across the country spent most of the day without service.
IDC's Harris said major outages are becoming more common among ISPs.
"People are becoming more used to outages. It's really sad, but all of the ISPs have really been having big outages," Harris said.
Still, Harris said that a service failure of the duration of AT&T's is out of the ordinary.
"Nationwide outage for a full day is unusual, especially for such a large carrier," Harris said.
He noted that such outages might account for the spike in subscriptions to low-or no-fee ISPs.
"One of the biggest reasons there are free ISPs is not because people don't want to give up the 20 bucks a month for their accounts, but more because they want a backup," Harris said.
Harris noted that almost half of the subscribers to ISPs such as NetZero rarely use the accounts, which are primarily a last resort.
Mark Goldston, CEO of NetZero, in California, said such outages can boost business.
"If you have even 60 per cent of all your traffic using your service and it goes down, that is a massive window of opportunity for another ISP," he said.