Original URL: http://www.theregister.co.uk/2006/06/26/winfs_axed/

Microsoft's future file system dies, again

WinFS cops it

By Andrew Orlowski

Posted in Software, 26th June 2006 15:51 GMT

Microsoft's most ambitious software plan - to base Windows on a native database - has died again. The feature was originally touted in 1991 for 'Cairo', which Microsoft then described as an object-oriented operating system, built on top of Windows NT. Cairo was sidelined as a result of Microsoft's focus on the internet, and the evaporation of the Apple/IBM Taligent OS. But the idea, reborn as WinFS, was revived in 2001 as one of the "three pillars" of Longhorn, now Windows Vista.

Now it looks as if Windows on a database won't take place until the next decade, and there are serious doubts it will ever happen at all.

WinFS Team's Quentin Clark wrote on Friday that Microsoft would not be releasing WinFS as a plug-in for Windows XP or Windows Vista.

"These changes do mean that we are not pursuing a separate delivery of WinFS, including the previously planned Beta 2 release," he wrote. "With most of our effort now working towards productizing mature aspects of the WinFS project into SQL and ADO.NET, we do not need to deliver a separate WinFS offering."

WinFS was already a depleted version of what Microsoft announced in 2001. Then, the idea was to base raw, native Windows I/O on an SQL database. Existing file systems, such as NTFS, would act as plug-ins. However this ambitious goal was first scaled back in 2003 abandoned in 2004, when Microsoft announced it would be built on top of NTFS after all. What we'd assumed was an acronym for "Windows File System" was actually the less committal "Windows Future Storage".

A year later WinFS was formally 'decoupled' from Longhorn - meaning it wouldn't ship at the same time as Longhorn itself. Instead it would be an "out-of-band add-on pack".

The ambition, if realized, certainly had its merits. Basic I/O semantics haven't changed for thirty years, despite Microsoft's attempts in the early 1990s to introduce abstraction (in the form of streams) to its Office files. As a consequence, systems conceived in the 1970s, such as Pick OS and IBM's AS/400 minicomputer can boast more advanced, database-like features.

It's partly as a consequence of this that today most data resides in proprietary formats - and Microsoft still fights open formats (and protocols) with a zeal. And while the world wide web provides a unified namespace, of sorts, it's one without transactional integrity or the other trappings of a well designed database.

Yet at the time news of WinFS first leaked out, it even prompted antitrust concerns. Read how we broke the news, the fears it raised, and how the BeOS architects anticipated the problems Microsoft eventually ran into.

So now, how can Microsoft vanquish Oracle, if it can't bundle a database with every copy of Windows? ®