Berkeley DB Family

Hi,

beforehand, sorry for my bad English...

I want to implement an RDF-Store using Berkeley DB. That means, I have a file with RDF-Triples (Subject, Predicate, Object) an want to store them in a Berkeley DB. Then I want to operate some Simple Graph Patterns to get some information about the data. My problem is, I do not know exactly, how to store my data in BDB... Since there are only key/data pairs, how can I represent the Information given by a triple?

My thought is to store the information in multiple tables, where one part is the key and the other two parts are data (for example: subject as key and a pair (predicate,object) as data). But I think this isn't a good approach...
How does it work with these secondary indices? can I just store a table with subject/predicate and link the object to this tuple (not only to the predicate, but to the combination subject/predicate)?

What would you suggest?

Thanks for your help,

Thomas