Matthew
Our university provides IBM Rational Rose Enterprise edition for internal use so I thought that using UML would be the best way ahead. Would people agree?
I would agree that it is, but not using tools. Pencil and paper is the way to go, especially in the beginning. That, of course, is a highly biased personal opinion. I just feel that modern tools are not up to it yet, it requires so much effort to get the diagram at the appropriate level of abstraction (and to look good on top of that!) that most people feel they're wasting their time and jump to code straight ahead. Which is not any good actually.
the data is extremely complex with nested, overlapping, and discontinuous heirarchies of objects, often in many-many relationships.
I would suggest to you a basic book on UML and modelling - my personal recommendation would be "Applying UML and Patterns" by Craig Larman. Chances are there are be several copies in your university's library. And also, although I haven't read it (yet ;)), I've been told that "Analysis Patterns" by Martin Fowler is a little gem when it comes to this area of development.
Note well that many of the most useful concepts in these books can be applied directly in Cach�, so don't limit yourself to look for the "data modelling" bits because that's usually pared down for relational systems. Instead, take a look at all of the structural patterns and concepts and try to figure out how to map them to Cach� (it's easy and we can help :)) Larman discusses some persistancy issues, you can also Google for Scott Ambler's papers on the topic. Though they would not be highly relevant to Cach�, I think they'd give you an excellent introduction to object persistence in general and some problems inherent to it.
Do come back with specific queries if you need to, there's lot of people from which you can learn here! (I'm still learning, so I'm not necessarily one of them :) )
HTH,
Ram�n
-- ZCacheLib - Open Source Extensions for Cach� http://www.zcachelib.org
