I love the post, and its works great , with the caveat on data size.
I have this working fine on 100M records, I want to get it to 5 Billion. Once I go above 200M records (av len 15 bytes) performance drops to unacceptable levels. Rebuilding the index takes forever, and doesn't actually seem to work properly once its rebuilt.
Is there any way to break up/partition a full text index?
Are there any guidelines on scaling FTS up?