The recent weeklong access foul-up with MSN Messenger left users angry and analysts questioning Microsoft's ability to effectively manage enterprise applications and systems under its .Net initiative.
From July 3 through July 10, up to one-third of subscribers worldwide were unable to access the free instant messaging service from MSN, Microsoft's Internet service provider unit.
The initial problem, according to Microsoft, stemmed from the failure of a disk controller on a database server; the backup server also failed. About one-third of MSN Messenger accounts were on that server, the company said.
While Microsoft estimates that there are 38 million accounts worldwide, messaging analyst Robert Mahowald at IDC in Framingham, Mass., said the number of users is probably closer to 30 million.
Microsoft wouldn't specify why it took eight days to restore service.
E-mail sent to Computerworld by irate MSN Messenger subscribers indicated a high level of user frustration. Joshua Lowe, a systems administrator at AT&T Corp., said he and a handful of colleagues use MSN Messenger to communicate between buildings at the company's Atlanta offices. It's the only free instant messaging service that works through his company's firewall, Lowe said.
"Microsoft itself has provided little to no timely or useful information regarding the status of the system," Lowe said.
Microsoft officials said they have learned from their mistakes and will be able to provide reliable service from now on.
"I feel very comfortable that this type of problem will never happen again," said MSN group product manager Bob Visse. "I would be shocked and amazed if this ever happened again."
But the initial failure and the lengthy restoration process could have implications for large corporations considering Microsoft's .Net initiative.
"I'm very concerned about HailStorm whether it's a good idea at all," said analyst Michael Sampson at Ferris Research Inc. in San Francisco, referring to the XML-based Web services plan associated with .Net.
"I'm very skeptical that they'll be able to pull it off, particularly in light of the eight-day outage," he said. "If Microsoft themselves can't get it right and reliable, then . . . this goes to prove that their products are not ready for prime-time use in the enterprise."
The reliability of .Net should be paramount, Mahowald said.
"You need to build a system of autoredundancy and more layers than Microsoft has done," he said. Pushing Web-based applications with .Net means that those applications will depend on the same kind of service Microsoft couldn't restore for a week, he noted.
"This seriously calls into question Microsoft's ability to adequately provision a reliable enough service for mission-critical enterprise use," Sampson said. "After what's happened with MSN Messenger, and the extremely poor way that Microsoft handled it from a communications perspective, unless they were willing to sign guaranteed service uptime with big financial penalties for nonperformance, I would counsel against building an enterprise business case around this technology."
Mahowald said the way Microsoft has chosen to develop and roll out .Net may be overly ambitious, so such problems should be expected. Rival IBM's piecemeal approach to building Web services, on the other hand, is more realistic, he said. IBM's narrower focus at least indicates that it's sensitive to enterprise needs, and it has already begun to address those, he said.