Datasheet
SYSTEM OVERVIEW
7
archical file structure found in FAT and NTFS to an object-oriented relational database structure,
where data is interwoven by tags and metadata.
Step back in time (if you would) to 1998 when the concept of an object-oriented file system was
part of the Cairo project (circa 1991–1995). Cairo was the Twilight Zone of Windows Server, and the
road to Cairo had many potholes. Cairo was abandoned, the object-oriented file system was put on
hold, and a less ambitious project called Windows NT 5.0 came into being. According to legend,
Windows NT 5.0 was the last major version of the Windows operating system that didn’t bear a
code name. NT 5.0 was branded as Windows 2000 Server.
The object-oriented file system was also a natural tool for the Component Object Model (COM)
(begun in 1993), as COM supports object-oriented distributed IPC. A new file system called
Storage+ was described (but never shipped) that supported COM. COM is the direct ancestor of a
slew of technologies, including Object Linking and Embedding (OLE), OLE Automation, ActiveX,
COM+, and Distributed COM. COM laid the groundwork for the current era’s technologies:
Microsoft .NET Framework and Web Services supported by the Windows Communication
Foundation (WCF).
Given this long history of development, it’s worth noting exactly why this technology has
garnered so much interest at Microsoft. As it stands now, data stored by applications running on
Windows can be “rich” in the sense that they contain complex data, but they aren’t easily integrated
with the data from other applications. This is a long-standing computer industry problem. A
hospital might have a database that stores text fields, memo fields (large text fields), number fields,
picture fields, even things commonly called BLOBs (Binary Large Objects) in a unified data store.
That database might know that one field stores phone numbers, another field stores the X-ray
images of its patients, and so on, but another application wouldn’t be able to have knowledge of
this metadata without a lot of additional programming.
Here’s where the object-oriented file system comes into play. Suppose the hospital wanted to do
a search for people of a certain age, who lived in a certain location, and whose X-ray indicated that
they had a condition of concern in common so that they could contact those people with a letter.
You can certainly do many, if not most of those things, in a database, but you couldn’t do all of that
from your operating system because the data is stored in proprietary or different formats, and
stored without the metadata necessary to tie it all together. But if the file system could do this, you
would achieve the following:
◆
Data independence of any application
◆
Integrated storage with a single data model
◆
Single instance storage where only one instance of the same object is stored
◆
Advanced search and data collection
◆
Expert systems and advanced data mining
So it’s clear why Microsoft continues to develop technology in this area, and why what they
learn becomes part of every major new data store such as Exchange, SQL Server, or Active Direc-
tory, that we see—just not, for the time being, the Windows Server file system.
The Child of .NET
Every version of the Windows operating system has at least one major initiative that it becomes
known for. With Windows 95 and Windows NT 3.5 and 4.0, the major advances were made in net-
working and internetworking. Windows 95 gave us Internet Explorer integrated into the operating
74593.book Page 7 Wednesday, January 9, 2008 4:11 PM