Hi Russel,

>I have an ASP application communicating with a VFP 8 DLL over the internet via IIS.
>
>A potentail Client has asked me:
>What is the maximum number of simultaneous requests that can be processed in this way without causing a problem. The server is P4, 2400 with 4GB running Windows 2000 Server.
>
>I need a realistic answer/guess if you have one.
>Please - No answers like - you will need to test it to find out.

In addition to Jos' comments, you also have to look at the way the vfp-code is called. AFAIR there was a simple perf test done by a company in England (mainly to get comparable numbers for their own server written in VFP), which showed a dramatic increase in responses if the vfp-object is instanced at the asp-app level and not with each call. The test page was very simple with a file based counter. They also measured other approaches to adressing vfp from the internet, and got base perfomances ranging between 20 and 700, AFAIR based on a Athlon 1500 or something like that.

There is a page on the west-wind wiki on performance hits which differnt methods of Rick Strahl's framework incur - done a few years ago, but you can get a feel for the possible numbers. Highly educational IMHO.

In other words, for vfp components having nearly no workload, you can expect on this machine, depending on your access method, 40 to 1000 Hits per second. With Asp (depending on the way the vfp component is called) this should give you 50 to 400 Hits per second. IOW: I don't think, that the overhead of calling your component will be the main factor for any project that doesn't have the budget to do some timing runs in the analytical phase and/or has the server already specified. (I couldn't resist <g>).

That still leaves the performance of your vfp-written components, and there Jos is also correct stating: "you will need to test it to find out."

These perf tests would best be done with the whole site, but you can probably get very realistic numbers for the vfp-parts by calling them yourself from vfp.

my 0.02 EUR

thomas

P.S.

One other link I found on my work-machine was
www.15seconds.com/issue/001009.Htm
which is not based on vfp at all - just to give you more data on how to conduct your on tests.