Question: .Net/Oracle application best design when building sql command parameters

Currently working on a web-service with Oracle10g as back-end. Here is my problem.

1. The table I need to populate has 100 columns. I have no control over the design of the table.

2. I wanted to use the "deriveparameters" method of OracleCommandBuilder, but found out that this is "costly" in terms of performance. What I initially planned to do was to derive the parameters from an open connection, close the connection, then maybe use the state management options of ASP.Net to house the parameters obtained but I don't know if this is possible.

I am unsure how big of a hit the performance will get if I continue using the "deriveparameters" method WITHOUT caching the parameters driven. If anyone can give me a link or article about the performance issues when using the said method, it would be greatly appreciated.

The environment is as follows:

-intranet
-around 2000 users
-users spread out in different parts of the country

I am not familiar where they will put the web pages of the application, the web service and the database. I don't know if they'll be putting it in different servers or on one server, but that's something we cannot control too, that's why I am looking for the best approach.

I just don't want to hard-code all that parameters and if, for instance, we need to change some properties/parameters, it will be a pain to maintain the code. It would be nice if someone here can point a way to automate the whole thing w/o the performance of the system taking a hit.

Also, is it possible to generate a class during runtime based on a table or a set of parameters from a stored procedure? If it is, can you point me to links on how to do it?

Thank you.