Re: Alternate solutions (Was: Re: NFS still has caching problem)

lilo (
Wed, 24 Jul 1996 07:32:43 -0500 (CDT)

On Mon, 22 Jul 1996, Theodore Ts'o wrote:

> why should we follow what sun does? We wrote our own filesystem
> because it is better than ufs. similarly we can still have NFS, no matter
> how broken it is and move forward at the same time by creating a better
> filesystem ( one that works rather than just research based; although we
> can pick the minds of people that are versed in it, there were some good
> concepts in the Sprite filesystem ).
> Because for a network filesystem, and indeed any client/server
> application, it's probably not worth our collective time to write a
> Linux-only solution. It won't be able to talk to all of the NFS servers
> that are running on other platforms.
> One of of the things which Microsoft finally learned was that
> solutions that depend on everyone running the same operating system,
> both clients and servers, just aren't going to cut it. That's why
> they've started embracing Internet protocols. And if Windows, with its
> huge user base, wasn't able to hack it, Linux certainly won't be able to
> win with a Linux-only networking filesystem solution.
> If someone wants to try, they are of course welcome --- it's a
> free country, and people can spend their time on whatever they want. I
> simply suggest that there are probably much better ways that a good
> kernel hacker might spend their time.

On the other hand, a solution developed on Linux using the free-software
model is not going to be a proprietary solution. Available source code can
help in acceptance.

If there is no good, non-proprietary solution available, there's nothing
wrong with developing a portable solution under Linux. If NFS simply won't
cut it, you design something that will. That's what RFC's are for....