File System Fandango

Microsoft has been steadily increasing its influence over standards. Sometimes it forces a new standard through shear monopolistic power, such as bundling Internet Explorer with Windows. These displays of raw power distract the public from far more subversive and far-reaching initiatives from Redmond. Through the .Net initiative and the Smart-tags feature of Office XP, Microsoft has been slowly laying the groundwork for a coup of unprecedented proportions. Borrowing a few ideas from Apple and other sources, Microsoft has found an unlikely weapon in its quest for computing dominance: a file system.

To understand how a file system could be so influential, it is helpful to consider the history of file systems. A "flat" file system is the simplest type of file system, where all of the files of the disk are in one list without directories. Many mainframes operate in this mode and can sort through tremendously long lists of files with amazing speed.

The next step in file system evolution was a baby step that started on IBM mainframes and migrated to MS-DOS. The three-letter file suffix used with IBM mainframes (and later DOS) was the first addition of meta-data to each file. Meta-data is descriptor data, or added data that describes the file. The three-letter suffix only allowed the barest of information to be appended to each file, denoting what "type" of file one was working with. This paucity is logical considering the constraints on disk space and the processing speed limitations of computers 30 years ago.

The next step in file system evolution was the addition of directories, or hierarchal storage, to the file system paradigm. As users added more files to their systems, lists became inefficient, because they could not be easily scanned and compared in the same way text files could be compared. MS-DOS contains directories and three-letter suffixes, but it was designed with the limitations of a floppy disk and a sixteen-bit processor in mind.

The first truly post-modern file system is the Macintosh's HFS (Heirarchical Filing System). Basically, every file was linked (via the resource fork) to meta-data that provided far more information about the file than could be provided by a three character descriptor. This added meta-data (resource fork) allows files to keep track of their creator applications, allows shortcuts (aliases) to follow the executable to which they were linked, and so forth. The implementation of forks makes the file system somewhat slower than comparable, less-intelligent file systems, but many find the tradeoffs acceptable.

It would appear that Microsoft has finally embraced meta-data big time. Soon, every file created by a Microsoft application on a local machine will have associated meta-data (i.e., Smart Tags). Third party support for Smart Tags may appear in the future. Microsoft's .Net initiative outlines standards for network based data as well. By wrapping everything in meta-data, Smart Tags on local PCs, and .Net over networks, Microsoft is laying the groundwork for its next step.

Microsoft plans to release a version of Windows with a SQL-based file system (SQL is a relational database program). If it does not appear in Windows 2002 (the successor to WindowsXP), it will appear in the subsequent release of Windows. Moving to a SQL-based system turns the file system into an intelligent database. By pre-wrapping everything in meta-data, Microsoft is preparing to create an ultimately searchable and possibly intelligent file system by adding an intelligent and extensible search engine (SQL).

This will firmly entrench SQL as the standard database and will be the first step towards eradicating competing companies like Oracle. Why use Oracle when SQL can integrate with the file system and handle data so much more intelligently? In many ways, it will have the same effect as integrating Internet Explorer: devastating any competition. Publishing a database driven website will be a simple function of the operating system.

For home and office users, the prospective benefits could be tremendous. Imagine working on a newsletter describing the latest release of Microsoft Office. You click on "insert graphic" and the OS brings up a dialogue box that includes not only a file tree, but also a list of Microsoft Office screen shots. The Smart Tags appended to the screen shots allow the SQL-based file system to make intelligent suggestions.

This also poses huge security issues. For example, will the Smart Tags be used to track your commercial buying habits, website visitation, or credit card numbers? Will Microsoft have access to your Smart Tags?

Microsoft is borrowing an old idea that was expounded upon by Apple and taking it to the next level. Throughout the early nineties, Apple worked on a number of exciting technologies that simply were ahead of their time. OpenDoc could have brought advanced functionality to Macs, however, through different means and with reduced performance. Taligent and Pink also could have transformed how users and files interact, but both projects were stillborn. Apple was loosing money on these projects, so it is understandable that they were discontinued.

However, Apple has gone too far in the other direction, away from innovation. Mac OS X is the ultimate "me too" operating system. It has all the features of Unix, with a dock and Display PDF - just like NeXT, which had the power of Unix with a dock and Display Postscript. It is easier to use, and Macs are about user experience, but Microsoft, while monopolistic and all around unpleasant, is producing technology after technology, from C# to .Net to Smart Tags. Hopefully now that Apple has a robust OS to base their innovations on, they will again focus at least in part on creating new technologies.

Apple has time to counter Microsoft's initiatives or adapt OS X to work with them. Microsoft rarely gets it right the first time, so the first iteration of their SQL based file system is likely to be horribly implemented. The fact that the file system is such an esoteric part of the operating system could allow Microsoft to slip under everyone's radar and would severely limit the scope of any lawsuit brought against the company. Hopefully, the SQL file system will act as a catalyst for competition, forcing Apple or others to create better alternative technologies to improve computing.

---

Chris Lozaga is a technical writer and has documented software for the IBM SP super computer and the AIX Operating System. He is no longer an IBM employee; this article represents his opinion and his opinion only. It is in no way indicative of the views of his employers, past or present.