I'm wondering... All documentation I've seen mention hdb as sibling of bdb with the only significant difference of being able to rename (sub)trees, and thus move them around.
Then there is one config difference: idlcachesize shoud be 4 x cachesize when using hdb. For bdb it can be way smaller than cachesize.
If the difference really is this small,
1) where is the real point in having two almost same back-ends?
2) why wouldn't everybody use hdb in the first place?
--Kari Mattsson
--On June 12, 2007 9:21:49 AM +0300 Kari Mattsson kari@trivore.com wrote:
I'm wondering... All documentation I've seen mention hdb as sibling of bdb with the only significant difference of being able to rename (sub)trees, and thus move them around.
Then there is one config difference: idlcachesize shoud be 4 x cachesize when using hdb. For bdb it can be way smaller than cachesize.
If the difference really is this small,
where is the real point in having two almost same back-ends?
why wouldn't everybody use hdb in the first place?
hdb was developed after BDB, so it went one release longer in the testing phase. Current plans are to phase out bdb in a future release (perhaps 2.5) and only have hdb available. ldbm was kept from 1.x->2.3, even though bdb was introduced in 2.1, etc. The main issue with hdb compared to bdb I see is the initial caching that needs to be done on large databases for it to get performant.
--Quanah
-- Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
openldap-software@openldap.org