John,

I believe that power-saving settings *can* affect performance. I've found that turning them off is more reliable and it is my standard procedure to do so (except on my laptop, where I leave certain battery-only settings active, and I have found erratic performance on it sometimes too that may be related, even though I am running with wall power).

You say that you notice that the mouse cursor on the server hesitates when updating is being done. That suggests that the server is also being used for stuff other than 'serving' the data to the users. You might want to try stopping such actions to see if that makes things better.
As well as on the user PCs you will also want to listen to the server to see if there might be an indication of SWAPping going on there too, and not just when table updates are happening on it. Maybe one of the other applications on the server is causing paging there.

I don't have experience with sustained network/application activity on a Win9x peer type network. I would guess, though, that your TRANSACTIONS should involve a minimum number of records before you end them or rollback. If you can have hundreds of records involved then I GUESS that could be a problem. If the users have some control on update/rollback decision, is it possible tht one of them is too cautious in their usage (holding on to too many before committing them)?

Do the users operate always in the same form or do they usually open/close forms often, depending on what they are doing? If constant open/close, then I think there is more chance for dangling references, especially because you say the trouble happens after 7-8 hours of usage. Does it make a difference for any one user experiencing the problem to reboot and restart the application? If it does then this is a stronger sign of dangling references.

As regards your child records problem...
You need to tell us what you know about the problem. Things like:
1) does it happen all the time or intermittently?
2) does it happen on user terminals generally or does it seem to be specific to one particular user/terminal?
3) are these records part of transactions too?
The most likely source of this kind of problem is bad programming, though no doubt there may be bugs in VFP that can cause this (just speculating. I have no specific bug in mund).

good luck
JimN


>Hi, Thank you for your reply.
>
>
>1) What is the normal speed for update 2 tables and 3-5 tables for getting info?
> WITHIN ONE SECOND
>
>2) Which of the following are you using...
> Pessimistic or optimistic locking?
> Row or table buffering?
> TABLE BUFFERING
>
>3) What RAM do the using PCs have installed?
> HYUNDAI/LGS, 64MB in client and 256MB in server
>
>4) What is the network operating system?
> USING WIN98 As OS in server with TCPIP protocol.
>
>5) Do you hear the hard drives running on the using PCs when the slowdown starts?
> Not sure. I will check it out later. However, the mouse cursor in server will hang for a while if updating from client machine.
>
>6) Are you using transaction processing?
> Yes.
>
>Your "my data may be not saved for my child table" seems to be a separate problem and we need more details about this too.
>What info do you need ?
>
>Does S.M.A.R.T., screen server setting and power saving in server machine effect the performance?
>
>
>Thank you in advance