RealTime IT News

Clearing a Path For Microsoft's Longhorn

Microsoft had little choice but to cut features from Longhorn, its next-generation operating system, analysts and developers said Friday.

Facing competition from Linux and a delayed product cycle, Redmond did what it had to. Longhorn may be the better for it.

On Friday, Microsoft said it would target broad availability of the Longhorn client in 2006 and would instead ship its WinFS sub storage feature after the release of Longhorn in 2006.

Prior to the announcement, the WinFS feature, which enables advanced searches on the desktop, was expected to ship with Longhorn; WinFS will be in beta testing when the Longhorn client becomes available.

Microsoft also plans to make key elements of its WinFX developer platform in Longhorn available for Windows XP and Windows Server 2003 in 2006. The developer platform will include Indigo, a messaging infrastructure based on a Web services-oriented architecture; and Avalon, the graphics presentation subsystem.

John Montgomery, Microsoft's director of the .Net development platform, told internetnews.com that feedback from customers and developers caused Redmond to take a hard look at what it could and should do with Longhorn's features.

Montgomery said Microsoft's leadership group had worked around the clock since it shipped the security-oriented Service Pack four weeks ago, and only made the decision last night on which features had to go.

Robert McLaws, the president and chief software architect of Interscape Technologies and a Microsoft developer partner, said some customers had told Microsoft they wouldn't even be installing the latest Service pack features for some time. The reason: security issues remain the top priority.

Another issue driving the decision was the dependencies on key new features that Longhorn would create for developers. "They found it was too difficult to program [for Windows] against WinFS," he explained. The feedback prompted the idea of de-coupling the WinFS system from Longhorn in order to help application developers program against the operating system more easily, without running into backwards compatibility issues on older platforms.

"If you think about it, that's key," he said. Part of de-coupling means you don't have cross dependency issues between development teams whose programs don't align with the core operating system, he added. "What Microsoft decided to do is ship a client that doesn't have those kind of dependencies, so you can get a regular developer experience and have it be what you've been expecting."

Redmond was also facing a serious product gap, said Illuminata senior analyst Gordon Haff. "With the competition they're facing, such as from Linux, Microsoft can't wait until 2009 for the next major operating system. Whether it comes out in '06 or '07, it needs to be a mid-decade release rather than -- heaven forbid -- an end-of-decade-release."

Haff said Longhorn was going to be "this wondrous city on the hill that would be everything to everybody and would incorporate all these new technologies and ways of doing things, while the schedule kept moving further and further out. I've been expecting Longhorn to be de-scoped for a while. But the feature cuts were absolutely necessary," he added. "Microsoft needed delivery dates."

Of course, the decision about what to keep and what to jettison in an operating system is complex.

Haff noted that the questions involve more than engineering. For one thing, Microsoft had to consider what incremental revenue might be tied to the various features. Another issue is the risk of the new: Features that are new in fundamental ways have a higher risk associated with them. Finally, when breaking new ground, "there's less historical data that will help them chart the path," he said.

Rob Helm, a Directions on Microsoft analyst, likes the new approach. "I think this is a good development," he said. "From time to time, people in the company have talked about delivering integrated waves of technology. This suggests Microsoft is trying to make it a little more modular and deliver wavelets earlier. It gives application developers a clear way to gradually move on to the new technology."

Microsoft hasn't decided how it will provide the final WinFS, Montgomery said. "WinFS is central to our vision of the way storage needs to work going forward," he said. "The hard tradeoff we had to make was, do we want to put all the other good stuff out there, or wait until we can do a full integration?"

The latest roadmap will help make security stronger, according to Helm. "Right now, XP is split on a different code base than Windows Server 2003. You can't create a patch that serves both, and that creates a more complicated base," he said. "I think this will help the plan to get the server and client delivered on the same code base as possible, which means it will be cheaper to deliver everything."

Illuminata's Haff said he wouldn't be surprised if Microsoft throws still more stuff off the truck. "I think Microsoft's ambitions for Longhorn were starting to exceed its development grasp, in terms of striking out in new technological directions in too many different axes." He said today's move reflects a newfound appreciation by Redmond's engineers on the fundamentals that need to be addressed.

Helm agreed. "I think at the Professional Developers Conference in 2003, Microsoft presented a bold vision and product. The bold vision is still there, but the products are more cautious and for corporations. I think that's a good thing, not a bad thing."

Interscape's McLaws said it was a gamble on Microsoft's part to "open the Kimono so early at PDC" and create higher expectations in the industry.

"It was a great thing, but I don't think the industry was ready for it. Microsoft had said the whole time, 'this is going to change.' But that ended up as a footnote."

Overall, he added, this signals that Microsoft is also clearing a path back to its roots as a platform company. "This is making it easier for guys like me to build applications that make it easier for everyone else."