>compare the values that you're seeing to the other box. Do any of them seem really out of line?
>
>I thought you only had to issue diskperf -Y to catch physical disk counters.
>
>-Mike

Well, I ended up going another way. The perfmon may be a nice tool, but I am unfamiliar with it, which is why I have to say "may be a nice tool", I just usually can get where I need to go on my own.

So anyway, I discovered at least that the network card was at fault, by doing these tests:

Retrieve New Machine to WS : SLOW
Retrieve Old Machine to WS : FAST
Retrieve New Machine to Self: VERY FAST
Retrieve New Machine to Old : SLOW
Retrieve Old Machine to Self: VERY FAST
Retrieve Old Machine to New : SLO

Which showed that every time the data had to leave the new server over the net, it choked. So we put in a new net card and that cured some of it, but not all. What I now have is basically this:

Test table, 25000 records, 500 byte fixed length (10 fields of 50 chars), populated entirely with foxpro's SYS(2015) function, which gives about 30% randomized data. In test after test, the old machine is delivering about 7200 recs/second, while the new one has gone from 1800 to 4000. So the net card was some of it, but not all of it.