On May 29, 11:51am, [EMAIL PROTECTED] wrote:
> The Micro$oft Dfs specifications in their Dfs white paper (as opposed
> to what is currently available for download) defines a first attempt at a
> distributed file system and really is a pale imitation of AFS and DFS
> -- but then it's free, right? There is no automatic replication,
> no replication whatsoever of the Dfs root, limited client caching,
> timeouts used for maintaining cache consistency, no RPC security, no
> encryption of data transfers, etcetera ad infinitum.
Right ... So Microsoft is giving us some breathing room. But it would be a
deadly mistake if we just sit there and breeeeeeathe, taking this market
opportunity lightly. Microsoft can easily put up a team of 100 developers
tomorrow, and start from the CMU Andrew File System or Coda. AFS will then be
the next Netscape, but without a single day of glory.
> Transarc, with the recent change in product status for AFS, is
> changing marketing strategy and initiatives for both AFS and DFS, and
> ISPs are just one targeted market (witness IBM's recently announced
> Nagano and WebSphere product bundles, which use AFS as the underlying
> file system: Nagano is targeted for Web sites and WebSphere to
> ISPs). I'm not entirely at liberty to discuss product futures, but
> suffice to say we are unwilling to concede the distributed file system
> world to either Sun or Micro$oft, and you will undoubtedly see more
> aggressive and interesting development and marketing of both AFS and DFS.
Mentioning Nagano or WebSphere is probably missing the point here. It is
a fine thing to approach the ISP's selling AFS as a Web site's backend or
as an enterprise file system, with the Olympics' success story.
But we are talking about Internet file serving/sharing here. We are talking
about emerging AFS to the front-end. We are talking about having ISP's
serving subscribers' files to allow users and their applications the freedom to
access files from the Internet with a genuine filesystem interface.
--
Shyh-Wei Luan
