Attain 3.10a & SQL2000 cluster performance

mandyk

Hi,
My client using Commerce Portal with 2 note Active-Passive MS clustering each note is 8 CPUs & 8GB of RAM.
They have 2 upfront web server running Commerce Server 2002 installed.
Approximately they have 30-40 regional concurrent users in Asia running Terminal Client and Client Local Network and I configured 3 NAS (2 NAS for job & reply and another one for SYNCH)
Current DB size is 8-9 GB and the peak load is about 2000 orders/hour from the front end users which are about 500 users.

There problem now I found of some message from NT event log that saying deadlocked on common tables such as item tracking line, reservation entry table & message queue table. Well it's NAS that become a victim of other process blocking a resource to long and causing inaccurate result of transaction. For example, some order created w/o lines and other things that really messy. :evil:
Well it is very bombshelf that even do Post-Ship of Transfer Order can cause other process waitting till it done and if there are others doing the same thing almost on the same time, it's like a queuing task.

well, I have another contraint that I put on the web application on message queue that only wait the reply message up to 20 seconds otherwise the user will get this message "No Reply From Navision".
I think you guys familiar with this message isn't ?
Meaning NAS can't be wait that long to have the answer, cos user will giving up and they make a lot of noise to support. :roll:

Well, I really need anyone idea or help.
I'm really think this is the max limit for Attain :shock:

Thank