On May 28, 2:56pm, James Roberts wrote:
> > I believe most people on this mailing list use and like AFS. It has
puzzled
> > me a long time why AFS has not taken off on the Internet. Other inferior
> > network file systems, such as NFS and Microsoft Dfs, are rushing to define
> > themselves as *the* Internet file system.
>
> Hasn't Microsoft DFS been mostly vaporware so far?
As far as I know, Microsoft Dfs has been vaporware in terms of server
manageability and scalability. But I think they are doing real work in the
area of performance (those AndX stuff), file locking, and security.
> Actually, I believe everything you mentioned is precisely IBM's vision for
> their DFS, which is in large part how their Web-DFS came about.
As much as I like to see DCE/DFS succeed (I was involved in the DCE design/
development in the early 90's), I think it would be a wrong weapon to be used
on the Internet file service market. DCE/DFS might be a good candidate for
intranets, where tight OS integrations in a more controlled environment may
warrant the price of the DCE complexity/baggages. For Internet, AFS is a much
better starting point, IMHO.
> CERFnet has an
> implementation similar to what you described--it's working very well from
> what I understand, but it's limited to just their ISP, and due to the
challenge
> of having to educate their users, their implementation of DFS is still
largely
> on the backend. I think this is also the vision of The Open Group's "IT
> DialTone," if I'm not mistaken.
The CERFnet model is not similar to what I described if DFS was only in
the backend, if it is just limited to their ISP, and if users have to access
the files through browsers.
I don't think AFS has much time left before the door to the mass Internet
file serving market close on it. We need Windows 9x clients, and we need to
give them to home users free. Get some ISP's sign up with the AFS file serving
business, and let the ball start rolling ...
Disclaimer: I only speak for myself here, not for my company.
--
Shyh-Wei Luan
