Re: exch 2003 /3GB problem

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance



1. Can you give me some actual MS documentation on the /NoExecute switch
that you are wanting me to add. This is a 32bit OS
2. SAN performance is fine. We monitor this quarterly.
3. I have used other numbers in the past besides 3030 per MS documentation
and there doesn't appear to be any changes
4. Where is the actual counter that you are wanting me to use. I could not
find one called VM Block Size under memory or processes. Let me know. Thanks

"John Fullbright [MVP]" wrote:

1. If you're at windows 2003 SP1, also include /NoExecute=optout in the
boot.ini

2. The fact that you're on a SAN does not automatically mean that the
storage is sized properly. Check your physical disk sec/write counters to
make sure your response times average under 20ms with no spikes over 50ms
lasting more than a few seconds.

3. You may need to use a value loke 2970 instead of 3030 for userva.

4. One of the most important VM counters is largest VM block size, what is
the value of this?



"Carpadum" <Carpadum@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:37D716CC-85CA-44C8-A231-FA0BD633024F@xxxxxxxxxxxxxxxx
We have Windows Server 2003 Enterprise edt. w/ 2 proc and 4 Gig of memory.
Exchange is also Enterprise Edt. and it is configured w/ 4 Storage Groups
and
each storage group has a min. of 2 Mailbox stores. The total database
size
of all these groups is around 65Gig. I have entered the following in the
boot.ini
[boot loader]
timeout=30
default=multi(0)disk(0)rdisk(0)partition(2)\WINDOWS
[operating systems]
multi(0)disk(0)rdisk(0)partition(2)\WINDOWS="Windows Server 2003,
Enterprise" /fastdetect /3GB /Userva=3030

(note the last entry is all on one line in the boot file)
I have also done the recommended reg edits to
HeapDeCommitFreeBlockThreshold
setting

When I do a perfmon on the Store.exe / virtual Bytes, the value is around
100 on the graph w/ .0000010 as a scale. When I look at the Free System
Page
Table Entries, the value is 2200 w/ .0100 scale. This seems way off to
me!

I am concerned that something is not running correctly. Why would the
store
process never use more than 1,100,000K of memory?(there is almost 2 gig
free
on te server) Why would there be so much free memory on a database of
this
size? The data lives on a SAN so it is a very fast machine, I just feel
that
memory issues are holding it up.



.



Relevant Pages