hashMap vs treeMap: on add 4 times slower and 3 times bigger on file system. It seems isn't ok? about jdbm3 HOT 3 CLOSED

Ok. It seems, common implementation issues. Dbm2 have same space occupation and perfomance.

On my data TreeMap a bit better on write in dbm3. But read in dbm2 still faster on 20-30%

from jdbm3.

Hi,
in short TreeMap has more advanced implementation in JDBM. For example it uses delta key compression which could explain smaller size.

Secondly I changed lot of parameters (page size, node size) and I did not do 'calibration' yet. So hash tree may be slower bcos its sizes are wrongly set.

from jdbm3.

BTree is much faster with linear inserts. HTree is faster with random access. So work as designed.

from jdbm3.