Hi,
I've inserted 100000 random boxes into a table (page file size about 8M) and
created a rtree index on it. Indexing needed a few minutes and finally I got
a 200MB page file for the index. Can someone explain this to me?
BTW: I did the same with Oracle 8i Spatial and the index seems to be a
lot smaller. Selection time for geometric queries (non-empty
intersection with a box) is roughly the same - a few seconds for both
2DBMS, but I don't have exact statistics.
I can guess a possible explanation for the smaller index in Oracle:
coordinates are rounded according the given accuracy, which could result
in smaller rtrees, but this is just a guess.
regards,
Christopher