Hi Jim,

>>I suspect that to be more in the networking/rights area,
>>but I haven't nailedit yet: it can kill performance
>>in a LAN.

>Why is it that it can kill performance i a LAN?
>Seems that the FIRST use would be slow from what you describe, but all >others should be fine. How does this "kill performance on a LAN"? Or did >you mean for one user only?

Point one:
we usually develop not in alive LAN,
so these performance issues creep into the product,
and the test team can't measure performance
with specific situations defined.

I have seen instances, where the program takes
over 2 minutes to perform steps which take
3 to 5 seconds on a stand-alone machine as well
as 4 to 6 seconds on an empty LAN.

Since this difference is creating quite a stir
with larger customers, I guess I will get
a test net with typical componenets and a few
days time to set up a reproducable test scenario
in which the performance drops under current
program some time this year.

Point two:
In our framework tables/views are closed, if they
are not needed any more, and possibly reopened
later on. If each "use" takes more than a sec,
performance of opening an app of getting
to another part like "customer data"
which opens between 15 to 60 tables could became
a real drag. The strategy to counter is clear,
if that is the culprit in my app (create a
special "caching" data session) but memory
will have to be watched even more...

HTH

thomas