Thanks David and Jonathan!

Yes rows doesn't have a name, I'm just using the word name for anything,
like cluster name,
table name, row name etc, that is my bad.

Yes, I did change two things, that was probably stupid, but the reason for
the second change
is space efficiency.

You are totally right that I'm choosing between scalability and performance
with the different
structures. What I really want to do is to just store indices in rows with a
composite key and
do range queries. Jonathan has firmly steered me away from this approach for
now in regards
to performance.

Thanks a lot!

Reply via email to