I'm of the opinion that we need to deal with it the other way, so
double-paging
_can't_ happen. If the Linux instance's kernel can be told that no
more than,
say, 5% of the memory can be used for buffer cache (I've been
perusing the
kernel source and I _think_ there's a way to do this), preferably via
a /proc
entry, then the instance can have no paging space known w/i Linux and
can
then be a fairly big (memory-wise) instance of 512MB or even 1G,
letting VM
itself handle paging requirements- once.
Mind you, this isn't real simple, but it cuts down on paging stuff in
from VM
just to decide we can turn it around for some other use.
And I date back to the days of XENIX 1.0 (Unix v7) and Unix
System-III on
teeny (896K) boxes where tuning the buffer pool was something you
looked
at closely.
--------------------
John R. Campbell, Speaker to Machines (GNUrd) {813-356|697}-5322
"Will Work for CLAIM Codes"
IBM Certified: IBM AIX 4.3 System Administration, System Support
"Post, Mark K" <[EMAIL PROTECTED]>@VM.MARIST.EDU> on 04/17/2002 01:44:18 PM
Please respond to Linux on 390 Port <[EMAIL PROTECTED]>
Sent by: Linux on 390 Port <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
cc:
Subject: Re: Linux cache
The only way I know of without tweaking kernel source is to put "pressure"
on Linux's storage. I.e., use V-disk for your Linux swap, keep your
virtual
machine size small, and the amount of storage allocated for buffers and
caching will be reduced.
Mark Post
-----Original Message-----
From: Mark D Pace [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, April 17, 2002 1:33 PM
To: [EMAIL PROTECTED]
Subject: Linux cache
I should have asked this at WAVV.
One suggestion for improved sharing of resources was to use VM MDC to cache
instead of Linux. I have been unable to find how to limit caching in
Linux. Any know how to accomplish this?
Thanks very much.
Mark D Pace
Senior Systems Engineer
Mainline Information Systems
1700 Summit Lake Drive
Tallahassee, FL. 32311
[EMAIL PROTECTED]
Phone: 850-219-5184
Fax: 850-219-5050
http://www.mainline.com
