Hi Byron and Eva,
if it is so (like Byron suggested), then I'll be happy to also do my little part in being unhelpful ;-)
In our company, the architecure is three-tier:
- First comes a qvw without GUI, loading data from the Database and saving qvd_files.
=> Then comes a "DataModel" that loads all the qvd_files and builds a data_model
=> That is then loaded (BINARY) into (most of) the apps.
as far as limits - so far I am conscious of only one real limit - well, rather a drawback, not an absolute limit - the size. A data_model of that kind can get huge and especially if people are connected via VPN and open an app using it, that might turn out troublesome.
You can, however - and we have tried that - create several different smaller DataModels, for just a few apps.
One thing we thought of trying, but haven't succeeded with yet, is
=> making a small DataModel that is sufficient for some apps and then
=> loading it BINARY into a secondary bigger DataModel and adding tables as per requirement for other apps.
Sorry I was away!! great to see answers !!Ok, I will try to explain clearly.
My Goal is to be able to load any new source just with a parameter file update.
The script will load a metamodel. The metamodel sctrucure comes from the script, metamodel data comes from the parameter file.
After the script will load all the data.
I did that for a small project. I was working despite I had some conceptualization issues. and It just adapt to the project, not enough general to be usefull.
My points: could we do all operations on this way? do we need to process operations in a special order? if yes, how to find the best order to optimize it? any other limitations?