As part of uni course I'm carrying out some consultancy work. The company 

I'm working for is BIG and relies on information being 100% (more or less) 

correct and rapid time-to-market development of new initiatives. They 

currently have a large client/server tool which is built in-house. The 

system is run on AS/400 and Unix/Sybase. The problem is that the 

production/live system is approx 1 Terabyte in size and 4 replicated 

UAT/Test environments are kept (2 are essential I've been told for 

statements, specials ets and the other two are full UAT), which means it 

is cumbersome and expensive to maintain. The main problem is that testing 

becomes a problem due to the large amount of data and testing often over 

runs due to this (and they're fast running out of server space).



My group have been asked to assess risks, benefits, costs and savings 

associated with moving testing over to a cut down version. The business 

resists these change saying that problems would be identified using cut 

down datasets due to the complexity of data in the system. Now this is 

where the problem lies. Where do I start!!!



As far as I can see, this is a question of scalability and I'm assuming 

that testing the data shouldn't be a problem, if the code works on small 

amounts of data I presume it should work on larger amounts :)



While I appreciate that every scenario is different, if any one can shed 

any light on what i should be looking for, the kind of questions I should 

be asking, test plans for this kind scenario, who i should be talking 

to/interviewing or where i can find more info on this kind of issue (or 

even keywords for search!) or any kind of help at all, I would much 

appreciate the help.



many thanks in advance

kamruz