So, we are about to port H2 code to a have a noSQL store instead of on the filesystem for our use in a large system(though with trillions of smaller indexes).
When looking at lucene and H2, at first glance it almost looks like they both use one b-tree if you index 4 columns(say A, B, C, D), instead of 4 b-trees. I am a bit confused there as that would mean, I would need to query on A or this would break down, correct? or am I mistaken and there is actually 4 b-trees, and when I do a join, that would mean there could be 8 b-trees I need to deal with or something.
Is there any good articles on how this works in detail? or can someone recommend some good book on this subject?
(I was an electrical engineer in school so never had that database programming class :( kinda regret that but shouldn't be too hard to catch up).
thanks, Dean
