I've looked at Cache'. It is not the "cure all" that they claim. It has issues, including performance. In the end, we ruled out OODBMS.


>Yes! And this is a key why we choose Cache': There is NO O-R mapping in Cache' at all.
>I believe it will be interesting for you to look at their unified data architecture concept: http://www.intersystems.com/cache/technology/components/uda/index.html
>In short, ALL data is defined as persistent classes. Once defined each class automatically can be queryied (or projected) as SQL-table.
>
>This is NOT a slow SQL emulation over instantiated objects and this is NOT object mapping over relational tables.
>The data is actually stored in sparse arrays which are used directly for both object and SQL access.
>
>(In fact, Cache' also allows direct access to storage arrays, which one can use for fast bulk load, for example. This is something similar to our C programs that write DBFs as binary files while more simple as we don't need to deal with physical file format.)