How would _you_ do this in "standard" Unix?

[Preston Bannister]

We use diskless Unix workstations to support an application where,
among other things, we display high resolution images of scanned paper
documents.

For performance reasons we both prefetch images into memory as well as
cache recently viewed viewed images.  This can add up to several
megabytes of cached image data.  

We're trying to support a fairly large number of workstations on the
same network, and on a single fileserver.  Anything the increases the
load on either unnecessarily is to be avoided.

Overall, it would be better if the cached image data was never swapped
(paged) to disk.  If physical memory becomes tight, some of the cached
data should be selectively purged from memory.  What I'd like is to
avoid swapping of cached data, by application controlled purging of
selected cached data.

My question is this:  How would you approach this problem using
'standard' Unix memory management?  Is there a reasonable solution?

--
Preston L. Bannister
USENET	   :	hplabs!felix!preston
BIX	   :	plb
CompuServe :	71350,3505
GEnie      :	p.bannister
--
Preston L. Bannister
USENET	   :	hplabs!felix!preston
BIX	   :	plb
CompuServe :	71350,3505
GEnie      :	p.bannister