tuning request and join index

I've a request that is like:

select l1.bill_id
from file f1

   join file f2 on f1.letter_id = f2.letter_id    join letter l on f1.letter_id = l.letter_id

so, the file table and letter table are big ( > 100000 rows) and I need that execution.
Join between both file aliases is not on ids, so there is no FK but there is an index.

So, how can I optimize the request or the schema to have fast answer ? What type of index should I create and how for such case ?

The only solution that my chief would like to avoid the materialized views...

Thanks a lot for your help Received on Wed Sep 23 2009 - 05:04:32 CDT