>>>>Absolutely, positively, beyond any shadow of a doubt wrong, especially with a dual processor configuration.
>>>
>>>
>>>Is that your final answer?
>>>
>>>Peter
>>
>>
>>He didn't even use a life-line :D)
>>
>>Mark
>
>
>Is he good or what!
>
Not really, but I do know enough to see how to make it work.
VFP won't split its work across multiple processors, but it can ensure that one processor is always available to the OS. This meand that background overhead-related things that can happen in parallel with memory-intensive operations in VFP don't get bogged down because VFP sucks all the cycles it can in a single atomic VFP command. And if you run multiple VFP sessions, you can obviously spread the load around...
NT AS can divide the address space to make sure that the application space gets more of the RAM than the OS - where Win9x and NT normally split the address space 2GB for system and 2GB for applications, NT AS allocates it 3GB to application and 1GB to OS. This means that at least potentially, the application can make use of a larger block of memory.
Ensuring that you use a Slot 2 (Xeon) processor makes a huge difference as far as memory; the Slot 1 architecture has cache problems with >512MB of memory in the system. Slot 2 doesn't have this problem. VFP can be tuned to keep its appetite curbed enough to avoid the swap file, resulting in big performance advantages, and you can tweak the priority of processes under NT or Win2K safely and relatively easily. Mistakes are les likely to kill the whole thing, too.
RAID can give significantly better disk bandwidth than a single spindle - currently, drives are limited much more by the speed that data can move on and off the platters than by channel bandwidth, and SCSI can have I/O in progress on multipel targets at once, so that the channel is able to be put to better use. Latency is an issue, but synced spindles alleviate this. RAID 0 gives maximum storage and speed; RAID 5 trades off disk storage and aggregate speed for fault tolerance.
If the issue of sequential, forward I/O is an issue (FAT has an advantage in straight sequential reads) WinNT can use FAT16, and Win2K, FAT32, at the expense of giving up the advantages of NTFS (better random read and seek performance, higher security, better fault tolerance.)