Re: Paging file -- MIDRANGE-L

This may come down to "how do you define 'swap,'" but, no I believe that in
the conventional sense, AS/400 does not swap.

I *believe* that, at least by old definitions, a system that did swapping
would have to actually swap out (write out) program instrutions and other
constants to the swap file... it couldn't / wouldn't go back to the
original source and retrieve those constant parts because they couldn't be
identified as such.  (Even program code on most systems - not iSeries - can
self-modify, so there goes your constancy out the window.)

iSeries SLS *IS* aware of constants, and in fact will retrieve them from
their original source if needed again - if the pages became removed due to
age.

Now I can picture you cringing from the efficiency of this, but now for my
second point:

How good is a swap file's use of disk heads?  If you have, say, forty disks
on your system, and you create a swap file, are you careful to allocate
equal parts from each of your disk drives, or are you going to busy one or
two of the disks with the swap space?  iSeries uses (read "takes advantage
of") all heads / arms on the system, providing a far better method of
storage (sans human effort) than any other system I've seen.

HTH
Dennis





James Rich <james@eaerich.com>@midrange.com on 10/29/2002 12:48:45 PM

Please respond to midrange-l@midrange.com

Sent by:    midrange-l-admin@midrange.com


To:    midrange-l@midrange.com
cc:
Subject:    Re: Paging file


On Tue, 29 Oct 2002, Dennis Lovelady wrote:

> Huh?

As I said, I don't know for sure - I'm just saying what seems apparent
based on the discussion so far.

> And this is circumvented on other systems?

Yes.  On my linux box memory is paged out to a partition of disk reserved
specifically for it (a swap partition).  So swapping out/in should go
relatively fast as compared to a random read/write.

> Anyway, I believe there's a flaw in your thinking.  If the data needs to
be
> rewritten, there's no guarantee that it'll go back to "the same place" on
> the disk.  In fact, if I understand it correctly (and I thought I did),
it
> likely won't go back to the same spot.  If it hasn't changed, it won't be
> rewritten.  If it has changed, it's subject to standard writing rules,
> again as I understand it.

My thinking may well be flawed - I don't claim any inside information on
this topic (I know a lot more about linux memory systems than AS/400
memory systems).

The question really boils down to: does the AS/400 swap?  My boss thinks
he may have heard that it does.  One post said that all objects have a
single address, whether on disk or in main memory.  If an object only has
one address, can there be more than one copy of that object?  Can an
object exist both on disk and in memory?  If so, then to which copy does
the address point?  If there can only be one copy of an object, where does
that object go when memory pressure is high?  If the AS/400 swaps, which
on disk copy of an object is pointed to?  If an object in memory is
changed, but the user does not want to write those changes permanently,
and memory pressure is high so that main memory cannot hold the changed
objects, where do those changes go?

Now clearly a copy of objects is made.  When some program object is loaded
of copy of it is made (with its own unique address) and loaded into
memory.  So when memory pressure is high, where do these objects go on
disk?  Is there a swap file/partition?

James Rich

_______________________________________________
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@midrange.com
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/cgi-bin/listinfo/midrange-l
or email: MIDRANGE-L-request@midrange.com
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.