Can Microsoft weather its HailStorm?

Not only has Microsoft been struggling to restore full service, but on Thursday the company also shut down MSN Messenger as it restarted the network of servers that handle messaging traffic. That "reboot" failed to immediately fix the problem.

The outage, which began Tuesday, affects as many as 10 million people, or roughly one-third of MSN Messenger users. Initially, many people simply lost buddy lists of friends, but as Microsoft tackled the problem more aggressively, service collapsed completely for many of those users. MSN Messenger customers continued to report service problems Monday.

For Microsoft, the outage is a black eye as it puts more emphasis on instant messaging--and network reliability--as part of its forthcoming .Net software-as-a-service strategy, which includes an ambitious services plan called HailStorm.

The MSN Messenger outage "shows that Microsoft has some work to do before they're ready for large volumes of critical traffic," said Gartner analyst Michael Silver. "This doesn't exactly look good for .Net.

"Rebooting when there is a problem with product servers is something we see more of in small to midsize businesses," not with a large commercial service, Silver said. "With (MSN Messenger), you're talking about an architectural problem, where Microsoft is being forced to reboot their servers."

The Microsoft of today is built on one-time sales of computer software, but the company in the future wants to move into the more lucrative services market, turning its products--through .Net--into rentable services for business transactions and consumer commerce.

Microsoft is unveiling .Net in pieces and is retooling its Windows operating system and desktop applications for HailStorm, its initiative for delivering content and services to virtually any type of device, from PCs to handhelds to cell phones. Instant messaging is one of HailStorm's most important components and one that Microsoft is shaping for the delivery of business services and information through the forthcoming Windows XP, Xbox, Stinger phone and other products.

In the HailStorm plan, Microsoft becomes the conduit for a wide range of consumer information, from addresses and schedules to credit card numbers and personal documents. The plan has come under heavy scrutiny from privacy advocates concerned that Microsoft will gain control of consumer data. Microsoft insists that HailStorm customers will "own" their data, which will be maintained in a secure repository under supervision of a third-party hosting company.

Charles Fitzgerald, Microsoft's director of business development, said earlier this year that another plan under discussion would let people host their data locally on "smart cards" to increase security. Fitzgerald also said that Microsoft has no plans to mine, sell, target or publish people's data stored in HailStorm.

But the MSN Messenger problem is the latest in a history of security and reliability problems that could make consumers less willing to trust Microsoft's HailStorm plan. The company has recently battled security problems in its Web server and e-mail software. And it came under heavy criticism for a widespread outage of Hotmail, its free e-mail service, earlier this year.

Some components of HailStorm have already stoked controversy, even though the service isn't expected to launch until next year. Privacy groups have criticized Microsoft's Passport authentication service, which will be a major component of HailStorm. Passport is a single sign-in service that gives consumers a key for multiple Web sites. Microsoft in April revised the "terms of use" policy for its Passport service after criticism that the agreement gave the software behemoth excessive control of customers' communications.

Questions about quality of service
Microsoft hopes that with HailStorm, which is in large part a consumer service, it will entice customers to pay a monthly or yearly fee for instant access to their personal data. However, the MSN Messenger outage calls into question Microsoft's ability to reliably run a venture as wide-ranging as HailStorm, analysts said.

The HailStorm plan would involve thousands of servers located worldwide, Microsoft has stated in the past. Initially, the company would host those services internally, but company executives have said Microsoft plans to outsource server operations to third parties.

How Microsoft will manage all of those servers and guarantee access to HailStorm services is the big unknown, Silver said. "The question here is, why did this outage happen?" he said. "Microsoft really has to come clean about what happened, because people are going to be under the impression that if Microsoft can't manage this kind of service," they might not be up to the task of managing HailStorm and .Net.

Bob Visse, MSN group product manager, dismissed Silver's concerns. "This is not a drastic action," he said. "It was something we carefully considered before moving forward with taking the service temporarily offline for as short a time as possible, to help get MSN Messenger back up and running for all users as quickly as possible."

The systems are dedicated solely to MSN Messenger, he said.

"This outage is not indicative of Microsoft's ability to move forward with its .Net strategy," Visse said. "This was one isolated issue brought on by a series of extremely rare hardware failures."

He also emphasized that Microsoft is "taking lessons learned from this outage to improve the MSN Messenger service for our customers."

At the least, Microsoft will need to repair its customer relations. Many readers who contacted CNET News.com criticized the company's response to the MSN Messenger outage and the .Net and HailStorm plan.

"As if 10 million users were not enough, if Microsoft plans to do e-commerce using instant messaging to notify users when their product will arrive, if it's delayed, etc., and their service drops for six days, it would be a serious blow to Microsoft's wallet and any other company using their technology," said Ian Fiebig, an MSN Messenger customer in Ottawa. "If Microsoft plans to integrate instant messaging into many of its major products, such as Windows XP, it has a lot of tweaking ahead," he said.

"You have to wonder what is actually going on, what actually takes (more than) six days to fix," he said.

To make the transition to HailStorm and .Net, Microsoft must make broad infrastructure changes to support the computing and bandwidth load necessary to support those extended services, the company has said.

In an interview earlier this year, Microsoft's Fitzgerald explained that HailStorm will include "a services fabric infrastructure monitored by Microsoft in the near term and distributed later. We will outsource the infrastructure to a third party."

Fitzgerald would not identify the third-party companies that Microsoft will use for server hosting, but did say "we're already one of Exodus' largest customers." Exodus Communications is a Web hosting and services company. HailStorm services will be built on Windows 2000 and SQL Server, Fitzgerald said.

Third parties will play a crucial role in establishing the network of distributed servers responsible for delivering HailStorm and other Web or software services, Microsoft Chairman Bill Gates said in an interview last month.

"All of those things are being done with other people," he said. "The very protocols of the Internet will evolve for security and quality of service and richer caching. And so we are out talking with the Ciscos and the Akamais and Intel--you name it--for that level of stuff. They want to evolve their products, too, to work with Web services," Gates said.

Microsoft in January inked a deal with Akamai Technologies to provide backup servers to protect against network outages after a series of hacker attacks that crippled its Web sites.

Rebooting the network
But moving to a broad, distributed network capable of handling the load, particularly in a crisis, is essential for any business delivering Web or software services, said Patrick Verbruggen, a software architect working for a Belgian Internet service provider.

"It indeed makes me worried about what will happen when HailStorm becomes a reality," he said. "At this point in time (with the MSN Messenger problem), the only thing that happens is that we can't chat anymore. It's extremely annoying, but at least there's backup for something like this: telephone, e-mail, etc."

For years, network administrators have complained about the frequency with which they have had to reboot Windows desktops or servers compared with competing products, such as various flavors of Unix or Linux. While Microsoft has made great strides reducing the need to reboot, the drastic action of essentially restarting its MSN Messenger network reflects poorly on the company's larger .Net services strategy, say analysts and consumers.

"The fact that the firm is inadvertently shooting itself in the foot now," said Dan Yurman, an MSN Messenger user from Idaho Falls, Idaho, "raises fundamental questions about HailStorm and the .Net strategy, especially with first commercial release just a few months away."

Don Cooper, executive director for the Greater New Orleans Broadcasters Association, said: "I'm not looking to bash Microsoft over an outage--an outage can happen to anyone. But an outage that lasts seven days with no valid explanation really starts to make you think about .Net, and about Microsoft's plans for the Internet. What if this were the new Office software verification service that was down?"

Microsoft requires users of its new Office XP and Windows XP software to contact it to activate the software.

Verbruggen raised similar, "kinda scary" concerns. "What would happen when people are really depending on HailStorm for business and personal use?" he asked. "Suppose I register for a number of paying services upon which some of my business depends, and I can't reach these services for a whole week. Even worse: Suppose I wake up one morning to see that Hailstorm has lost all trace of my registrations."