>>>IMHO, I don't think you can do a fair comparison because you are dealing with 2 very different types of databases. VFP will be faster (and much easier to deal with) if you retrieve and manage data the way you have to in Oracle (unless you have a crappy file server where VFP tables and DBC are stored versus a hummer of an Oracle server). You can not manage Oracle data the way you can in VFP. Oracle data will all be handled through remote views. If you have individual views on 2 Oracle tables, you can not relate the 2 views. You have to create a separate view that joins the 2 tables and selects the appropriate fields from each. This technique takes advantage of the indexes that exist on the Oracle side.
>>>
>>>In some cases, I have had better performance retrieving an appropriately limited dataset from an Oracle server than from local VFP tables on a local Novell file server. Of course I accessed Oracle via a dedicated TCP/IP T1 connection versus a local 4mbps Novell token ring to the VFP DB.
>>
>>Good point. I hadn't considered that when I proposed the test. I guess I would modify it to say given our set of requirements for front end functionality, an all VFP solution (properly design, but of course ...) that uses VFP data and VFP COM servers should beat an Oracle solution with any front end. But how to prove this? To a customer who is considering VFP versus Oracle?
>
>I think it is better said that VFP should be the front end, and the back end is VFP versus Oracle (or maybe even a combination of both). Oracle is just an RDBMS. Regardless of the backend, VFP is still the best front end. My apps that use Oracle as backend, have always had some local tables along with the remote views.
>
>Has the customer even considered the cost of Oracle? Much less the cost of an Oracle DBA, hardware for Oracle, maintenance? Oracle is extremely expensive (5 figures and counting) when compared to VFP.

Cost is a consideration and it is probably one reason why we are in the running for the work. Their concern is performance--they would like to see some metrics that compare a 3-tier Foxpro solution to an Oracle solution and show that the performace would be the same or better.