It is sort of true, the following is from Scott Hanselman's blog discussing msdn:
I figured there can't just be the
(loband) "switch" and I mentioned I
thought that having to hack the URL
was kind of wonky. Turns out that the
whole MSDN system isn't a bunch of
files on disk, but files in a database
with an ASP.NET Virtual Path Provider.
Tim Ewald wrote about how they did
this WAY back in February of 2005.
This was, at the time, kind of a
poor-man's ASP.NET Routing:
The normalized path points to a file
that does not exist on disk. Rather,
the page data is stored in the content
cache. The system uses a
VirtualPathProvider (VPP) to bridge
the gap between the two. In essence, a
VPP intercepts all of the ASP.NET
plumbing's requests for file streams
and gives you a chance to load them
from wherever you like. Every ASP.NET
app uses a default VPP that simply
maps to the file system. An MTPS-based
site registers a custom
VirtualPathProvider, which sits in
front of the default VPP, forming a
chain. The custom VPP uses the
DocumentInfo and ContentSet objects
that the HTTP module's
OnPreResolveRequestCache event handler
stored in HTTP context to load a topic
from the content cache and return it
as an .aspx file stream.
And if you want to know more about how MSDN is built have a look at the linked article from Tim Ewald