Microsoft kills the WinFS dream

Server admins' hopes fade as better Windows file system management goes the way of the dodo

It was what Microsoft described as one of the "pillars" of the Vista/Longhorn platform. It had geeks excited because it was more than just a smarter version of FAT32 or NTFS -- it combined relational data technology with local and network operating system concepts. It was ambitious. It was delayed. It was removed from Longhorn to be developed as its own platform. It was WinFS.

And now it’s dead.

Last week, a blog post on the WinFS Team blog announced that WinFS would no longer be developed as its own platform. This came just as many of us were expecting to see a Beta 2 release of the new file system show up in our mailboxes. Instead, the WinFSers have decided to take most mature portions of WinFS and integrate them with SQL Server and ADO.Net.

So as a stand-alone file system that would have dropped onto existing Longhorn installations, WinFS just did a face plant. Got snuffed. Pushed up a whole pile of daisies.

That’s not to say that developers won’t be happy. The relational aspects of WinFS are sure to make ADO.Net entities more powerful and certainly more slick when it comes to fast dev times. Similarly, SQL Server applications will get a serious boost, not only in terms of performance, but in what application developers will be able to do with a WinFS-bolstered SQL Server back end.

But for us server monkeys, we need to deal with the fact that we have no light at the end of the NTFS tunnel. There won't be a relational file system in Longhorn or any Windows server OS for the misty foreseeable future.

Many of us were looking forward to these features, simply from a server administrator’s perspective. WinFS would have let us create relationships between server data, using XML and a (supposedly) cool management front end. That means we could have built instant data relationships in much the same way that SharePoint will let us build instant team relationships. We also won’t be able to build that all-encompassing file system search utility that WinFS was promising.

WinFS’s relational capabilities would have allowed Microsoft’s search engine to become far more granular, detailed, and even faster into the bargain, including the capability of bundling proprietary data formats into search queries using metadata. No more.

And finally, WinFS would have seriously boosted Windows’ capability to perform flexible and speedy replication. True, Windows has replication features today, but they pale in comparison to what a fully optimized relational file system could have done, not just in terms of overall performance, but in terms of flexibility. The prospect of replicating various portions of the file system, possibly even with a rules engine on top of that, made many of us server admins positively swoon.

So aside from seeking grief counseling, what else can you do? Not much. Although there are third-party network file structures and replication engines out there, part of what made WinFS so attractive was the promise that it wouldn’t evoke all the headaches that a third-party bolt-on solution usually produces.

All we can really do is hope that Microsoft goes beyond the low-hanging fruit -- and by that I mean SQL and ADO. Integrating bits and pieces of WinFS on these platforms is a no-brainer for Redmond -- and not that difficult, considering where Beta 1 of WinFS was back at PDC 2005. But let’s hope they don’t stop there. Pieces of WinFS could also do well on Exchange, and especially on the new SharePoint with its hub-of-all-things-hubbable focus.

Maybe some genius programmers are out there right now building a working WinFS that Microsoft could just buy (hey, Google does it). In the meantime, I suppose there are worse things than NTFS.

Copyright © 2006 IDG Communications, Inc.

How to choose a low-code development platform