Well... if you've tested on 4 different LANs, the problem must be something they all have in common.

I think others have suggested NT tuning resources - however, I can't imagine all 4 NT LANs would have the exact same problem, that would manifest itself only with your application (i.e. other people haven't seen the same sort of thing with their applications).

So, at this point I'm tempted to think it's something to do with your app.

1. Have you relied on a commercial framework for your app, or have you "rolled your own" data handling routines?

2. Is there anything about the structure of your app that could be considered unusual e.g. one or more very large tables, hundreds or thousands of small files in a single folder, extensive use of OLE/COM, general fields, etc.?

Actually, one other thing to check - I don't know if it's been suggested already...

I had a client complaining that GoldMine was extremely slow on one PC. It was - it turned out that the virus scanner on that PC was set to scan ALL files, not just executables - made database access over the network brutally slow. Is it possible that local or server-based antivirus protection could be contributing? I know - you shouldn't see a lot of difference between 1 PC performance, and 2 PC performance if this was the case... just an idea.

>Al,
>
>I've tested this problem on at least 4 different LAN's that all show this problem. I'll look at the NIC drivers and make sure they are the latest.
>
>John
>
>>>Al,
>>>
>>>I tried just about every configuration there is and because it is being tested on several different networks, I think a bad ethernet card is out. I found an interesting cause and effect though...
>>>
>>>When 1 Win98 pc hits the app on the NT server, the cpu utilization on the server stays around 2-4 % during my testing. When I add an NT pc to the app and search again on the Win98 pc, the cpu % jumped to 22. When I run the test on NT pc cpu % did not change on the server. What could make the cpu work harder for the Win98 pc when another user is in the app than when the Win98 pc is in the app all by itself? If I get in the app with just NT pc's 3,4, or 5 pc's the test does not cause the cpu utilization to jump at all.
>>>
>>>Again, I really could use some help in figuring this out, MS is involved but the more help the merrier!
>>>
>>>John Brooks
>>>
>>A couple of way-out-in-left-field ideas...
>>
>>1. Are the NIC drivers in all the PCs (but especially the NT ones, including the server) EXACTLY correct? That is, are the NIC manufacturer's drivers (latest versions) being used, as opposed to "NE2000 compatible" or the like?
>>
>>2. If you're using PCI NICs, are any set up to share interrupts with any other card or device? Again, this applies to the server as well.
>>
>>A third left-field idea: do you have the opportunity to install and test your app on a different NT4 network? One that has been set up for a while and is known to be stable? That would at least determine if it's NT4 networks in general, or something about that network in particular.