Re: eliminating the swap file
- From: John John - MVP <audetweld@xxxxxxxxxxx>
- Date: Thu, 06 Aug 2009 13:45:50 -0300
BillW50 wrote:
In news:%23x01z5oFKHA.3816@xxxxxxxxxxxxxxxxxxxx,
John John - MVP typed on Thu, 06 Aug 2009 09:07:46 -0300:
Tracey wrote:I have a naive question: Is it possible to configure Windows 2000Windows 2000 does not run without a pagefile, if you try to disable
Pro to run without a swap file or move/redirect the swap file to ram
(C: is a FAT32 partition if it matters)?
the pagefile the system will give you a nag screen and create a 20MB
temporary pagefile (temppf.sys) in the System32 folder, the file will
be deleted the next time the system boots.
That could be John, I wouldn't know. All I know is creating a 20MB pagefile and throwing into a RAMDisk seems to keep Windows 2000 happy enough with only one nag screen at boot. Even increasing the swapfile over 200MB doesn't change this.
Possibly that is because the pagefile is created by the Session Manager well before the RAM disk is created (by which ever application you use to create the RAM disk), ergo the Session Manager has no place to create the pagefile when the system is booting, so it creates a temporary pagefile in the System32 folder and throws up the nag screen.
You can move the pagefile to any other disk or drive but you will have
to carefully consider the move, moving the pagefile to another
partition on the same disk as Windows or on a disk that is in a slave
relationship to the Windows disk may result in a performance hit and
is usually not a good idea.
Well the believers in defragging help improves performance (I never seen any noticeable improvements myself since IDE drives came along which replaced MFM drives) believe keeping the pagefile off on another partition help keeps fragmentation in better check. Thus improves performance.
So theoretically keeping the pagefile on another partition or another drive that shares the same data path (i.e. master and slave), and/or plus having the head having to move far to another partition should decrease performance.
Although Microsoft and the hardware manufactures knows about these bottlenecks and has taken steps to prevent these problems. Thus we have write delays and caches to solve these problems. So in the real world with modern hardware and enough RAM, I am not sure you would ever see a problem.
In the real world a busy Windows 2000 system with a pagefile on a slaved IDE disk will usually result in poor performance if the system needs to page to the disk, data can only flow to one disk at a time on an IDE controller so one disk will have to wait while read and writes are being performed on the other one, if the system doesn't use the pagefile or if it isn't heavily used then the performance hit may be negligible. Nonetheless, moving the pagefile to another partition on the same disk or to a slaved disk with the system is generally bad practice and it should be avoided if at all possible. This article may help:
http://support.microsoft.com/kb/197379/
Configuring paging files for optimization and recovery in Windows Server 2003, in Windows 2000, and in Windows NT
I only ask because the dialog box states that the minimum allowableWhat kind of applications are you running? Quite possibly the RAM
swap file size is 2MB (as opposed to 0MB which is what I was
thinking). This may be a {very very} bad idea (due to functional design), but I
just thought I would ask (see issue below).
Any insight on this subject will be greatly appreciated.
Thanks, Tracey
I am running with 2GB of ram (which appears to be only half used
with a swap file).
isn't being used because there is no need for it.
I was thinking that the computer could/should run faster/fastest if IThat is almost always a very bad idea! This is an idea that is often
upgraded ram to 4 GB and eliminated the swap file so all "memory"
activity would actually take place in ram.
advanced by well meaning individuals who unfortunately don't usually
understand the Windows NT memory management mechanism so they view the
pagefile as an evil presence. Few of them understand how the private
2GB flat address space that each process has available is mapped, they
fail to realize that without a pagefile the processes address space
can actually cause a very large amount of RAM to go completely unused
and wasted.
John
I disagree! Running Windows XP with only 512MB or less and Windows with 256MB or less RAM without a pagefile is almost always a bad idea. And I have experimented a lot without or with a very small pagefile on both XP and 2000. There are three big reasons why I'm interested in running without one.
In your situation with tiny SSD drives there may be some advantages not using a pagefile but for others with regular hard drives with plenty of space there is almost never any advantages in running without a pagefile. You should also keep in mind that the Windows XP memory manager handles this situation differently than Windows 2000. We sometimes hear of some who seem to experience no problems when running without a pagefile but we also hear plenty from some who have tried it and who have experienced problems, the pagefile hurts nothing and except for few special exceptions removing it is almost always a very bad idea.
Read the whole article and comments here: http://www.codinghorror.com/blog/archives/000422.html and you will see that the author came to the same conclusion after running his own tests, running without a pagefile provided no gain whatsoever and ended up causing him problems.
John
.
- References:
- Re: eliminating the swap file
- From: John John - MVP
- Re: eliminating the swap file
- From: BillW50
- Re: eliminating the swap file
- Prev by Date: Re: server reboots periodically - BugCheck 50
- Next by Date: Re: server reboots periodically - BugCheck 50
- Previous by thread: Re: eliminating the swap file
- Next by thread: Uninstall 2000 after XP installed over it?
- Index(es):
Relevant Pages
|
Loading
