Process time parameters

Hi!

A client wishes to use a fully normalised data structure, with only 2 or 3 columns per table. For example he has one table called 'Object' and another called 'Property'. Each object may have one or more properties, etc. His reason for wanting to use this very narrow structure is that he wants to secure interoperability between a large number of very different systems. He has decided that messaging is unsatisfactory as it requires each existing system to be amended to issue new messages when another is added to the set and so intends to use a common shared data base to act as a bulletin board between all those many systems. Because the problem structure of the systems is very different the 'common structure' that they all share is normalised to Hell and back. Some of the problem structures would be best supported by very wide tables with many columns.

The above description could excite a whole load of comments but the decisions have been made and can only be altered in the light of hard fact. [Maybe not even then.] What I need is a set of experimental figures where one variable - such as number of joins - has been altered at a time and comparative response times measured.

Can anyone suggest where I might obtain such figures? Does anyone know of any research on these lines?

Thank you for your help.

Ian

-- 
Ian Reid