By Stephen H. Wildstrom
The signs of trouble surrounding Microsoft's massive Longhorn development project have been building for months. But on Aug. 27 the software giant bowed to the inevitable. Rather than accept another long delay in the next version of Windows, Microsoft (MSFT ) announced that the new operating system would be released by the second half of 2006, but in attenuated form. The main casualty is a new way of storing data that's designed to make information much easier to find on PCs and networks.
The decision to scale back the ambitions of Longhorn stemmed from a complex series of events. First, what started out as a routine effort to package a large assortment of bug fixes and security patches turned into a project to overhaul the leaky security of Windows XP. The result: Windows XP Service Pack 2 was finished on Aug. 5, but only after hefty redeployment of programmer resources from Longhorn to XP.
MANY MORE MONTHS.
A second, more positive, development was the discovery of the superior performance of another project, Windows Server 2003, which is scheduled for release next year. "We started building Longhorn on the XP code base," says Greg Sullivan, Microsoft lead product manager for Windows clients. "But we found we were getting unbelievable reliability on Server 2003 SP1. We were seeing six nines [99.9999%] of reliability. So we realized it would behoove us to move Longhorn onto the 2003 code base."
The problem with the decision was that it added many more months of delay on Longhorn -- unless something drastic was done. "We saw that this thing was not shipping in 2006," says Sullivan. "Maybe it would in 2007." So Microsoft decided to drop the most complex and difficult part, a new file storage system called WinFS, out of Longhorn to shoot for a release deadline in the second half of 2006. The WinFS project is still alive, but its timing is uncertain.
The two other major components of Longhorn are a new messaging and communications system called Indigo and a new user interface called Avalon. Indigo, an overhaul and expansion of Microsoft's .NET Web services package, is the furthest along. It may end up being included in Server 2003 SP2 next year.
Avalon's status is murkier. Microsoft's announcement talked of making the new user interface, which will offer a variety of 3-D effects on high-performance computers and a simpler version for less powerful PCs, available for Windows XP, but it's not clear how that would be done.
Also unclear is the timing of a version of Windows XP designed for Advanced Micro Devices'(AMD ) Athlon 64 processor and Intel's(INTC ) forthcoming EM64T series processor, a 64-bit version of the Pentium 4. XP 64 had been scheduled for release this fall but was delayed until the first quarter of next year, and there are signs it may slip further.
In its Aug. 27 announcement, Microsoft also said some Longhorn features would be added to Windows XP. But don't get excited about seeing this anytime soon. Sullivan explains that the goal is to give programmers tools that will ease the transition to Longhorn by letting them write code that will run on XP while using routines, called application programing interfaces, or APIs, designed for the new Avalon user interface and the Indigo messaging system. This will make life easier for developers, but users will see little or no difference.
Since hardly anyone believed Microsoft's earlier claim that Longhorn would ship in the first half of 2006, the official move to the second half is not of great consequence. And what effect will the dropping of WinFS have on computer makers or independent software vendors? In one sense, the vendors' lives will be simplified, because Microsoft now plans to ship client and server versions of WinFS at the same time, whenever that is. Sullivan said many vendors were unhappy about the original plan to have the client and server versions separated by as much as a year, because it complicated their development plans.
The challenge for Microsoft will be to create a compelling case for going to Longhorn now that one of its most interesting features has been removed. But the marketing folks in Redmond, Wash., have a good two years to figure that one out.
A more serious problem is the embarrassment of having to scale back a long overdue project. The travails of Longhorn, already nicknamed Shorthorn in its attenuated form, recalled Microsoft's struggle with an overhaul of Windows code-named Cairo. The project occupied much of the company's energy and resources in the mid-1990s, only to be abandoned in a half-finished state.
Adding to the discomfort is the fact that after giving up the chief executive's job, Microsoft Chairman William H. Gates III named himself "chief software architect" and took personal command of the Longhorn effort. The decision to lower ambitions to avoid even greater delays was probably sound, but Longhorn is going to smart for a long time in Redmond.
Wildstrom is Technology & You columnist for BusinessWeek