Summary: The Web File System: // Filelike Access to the Web
Atul Adya, Joseph Banks, Jim Napier, Jordan Slott, and H.B. Weinberg.
May 20, 1995
The Web File System (WFS) is a file system interface to the World Wide Web. This interface helps
leverage existing investment in file system based tools by allowing them to access the Web as if it were
a file system. We do not propose the file system interface as an alternative to the existing graphical
interfaces; these will remain the preferred method of access for humans. Instead, WFS speaks to a
disenfranchised community --- our old friends like find, grep, csh and perl that until now were
largely barred from the Web. We also believe that WFS will make it easier to develop new Websavvy
applications by shielding programmers from the details of Web protocols.
Our aim is to provide a transparent system --- one that looks like a file server to its clients and like
a Web client to the Web servers. Our approach is to implement WFS as a userlevel network file server
using the NFS protocol . This implementation strategy is similar to that used in other systems that
provide filelike access to systems that are not exactly file systems, e.g., Alex (ftp access), OdeFs 
(access to an objectoriented database) and Semantic File System . The advantages of using the NFS
protocol are that many machines already use it, no modifications to the client's kernel are required, and
multiple client machines can connect to one NFS server.
In developing WFS, we were faced with many design choices including: what pathname maps to a
given URL; when to access the Web; how to cache data. Some choices were guided purely by aesthetics,