Unix RAM disks

[mike camann]

Excuse me if I'm being a bit slow-witted, and God help me if I draw the
flame attention THIS thread has recently generated, but wouldn't a
"true" RAM disk run counter to the UNIX file system sharing
philosophy?  Since all file systems are owned by someone, with strict
permission testing, wouldn't the creation of a file system in RAM
simply wall off that portion of RAM from other users?  Not an ideal
situation on multi-user machines, where user X's RAM disk suddenly
makes sufficient memory unavailable to user Y's application, or at the
very least forces user Y's processes into disk-swap slow-motion (I
guess a couple of megabytes here or there probably wouldn't make *too*
much difference, unless lots of users did it simultaneously).  Unless,
of course, the permissions were set wide open, but that's hardly a
desirable situation either.  There is still the problem of needing to
have root access in order to mount file-systems, as well, making it
likely that any system-wide RAM disks would simply be partitioned a

I suspect that Linux is a special case which is still rather closely
associated with single-usership on traditionally single-user machines.

Just my thoughts-- not meant to draw fire....

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Michael Camann                  camann at dial.pick.uga.edu
Department of Entomology        mcamann at sparc.ecology.uga.edu
University of Georgia           (706) 542-2303 voice
Athens, GA 30602                (706) 542-2640 FAX

After August 1, 1995:

Department of Biology           camann at lvc.edu
Lebanon Valley College
Annville, PA 17003

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%