>Robin,
>
>Here's something which ought to generate some contraversy - I think a case can easily be made that FRAGMENTATION IS GOOD in a true multitasking environment!
>
>I have never worked with NT Server, but I do believe that it is a true multitasking system. That being the case, and assuming that multitasking *IS* actually going on, then large files which are fragmented means that there is less "other" data to traverse to get at the data of interest. This should add up to be relevant pretty quickly.
>In fact I think MS ought to give us a way to "control" fragmentation better, just for such circumstances, as well as give us a "FRAGMENT" command where we could have some control over fragmentation of specific files and volumes.
>
>All this, as stated up front, for multitasking only.
>
>Cheers,
>Jim N
>
>>>We have a VFP application running on NT LAN. Sometimes it's slow, especially while working with big tables.
>>>How large are changes necessary for migration to C/S technology with
>>>SQL-server ?
>>>TIA
>>If you mean that you have the VFP files on a shared folder ( directory) on an NT server, you might consider optimizing your NT server environment. Specifically, the biggest hit could be: Fragmentation of NTFS partitions. There are several programs available from non-MS sources that address this fragmentation problem: Symantec's Norton Utilities, and Executive Software's Diskkeeper. We use Diskkeeper -- latest version will even defragment NT directories! Other considerations, include what else you are running on the NT server itself -- are you treating it like another workstation and just a repository for your VFP data as well, you may not be happy with the attention that NT pays to its own local keyboard and display. Several of our sites have upgraded the LAN hardware from 10 base T to 100 base T with some, but not staggering performance improvement. I can't take the time or space to indicate all the reasons you may not be satisfied with the performance you are experiencing on
>>NT, but without even knowing what you are trying to do the forementioned steps have helped us.
Controversy, or perhaps thinking of a different mthod of disk memory allocation / mapping -- NTFS ( from what I read in the Executive Software monthly bulletins ) involves a lot ( and I mean a lot ) of forward / backward chaining of data ( also places formerly occupied by data ), in fact from those same sources, one of the highest priority processes in all of NT is maintaining ( and periodically checking ) pointers. Makes for reasonably intact and physically secure data; however, memories of the old NOVELL 2.x Compsurf process -- which could take days on 100 megabyte drives. Finally, I would assume that your reference to 'good fragmentation' is the locality of reference to most recently added data? This would be nice if the OS wouldn't insist on referencing the physical data by forward chaining through every discreet segment of data between where the file started and where you want to be...