Dear All,
For the backend type for all the replication examples I'm doing just now, should we use bdb or hdb?
I notice that in slapd.conf.default we are using bdb.
Thoughts?
Gavin.
Gavin Henry wrote:
Dear All,
For the backend type for all the replication examples I'm doing just now, should we use bdb or hdb?
I notice that in slapd.conf.default we are using bdb.
I'm in no hurry to deprecate back-bdb, and back-hdb's initial performance is heavily dependent on priming its IDL cache. It's a tossup - back-bdb with no subtree renames but less cache sensitivity, or back-hdb with full directory semantics and more cache sensitivity. I suspect we'll be carrying both for a long time, because of this tradeoff.
Thoughts?
Gavin.
<quote who="Howard Chu">
Gavin Henry wrote:
Dear All,
For the backend type for all the replication examples I'm doing just now, should we use bdb or hdb?
I notice that in slapd.conf.default we are using bdb.
I'm in no hurry to deprecate back-bdb, and back-hdb's initial performance is heavily dependent on priming its IDL cache. It's a tossup - back-bdb with no subtree renames but less cache sensitivity, or back-hdb with full directory semantics and more cache sensitivity. I suspect we'll be carrying both for a long time, because of this tradeoff.
Understood. I'll stick with back-bdb for the examples, unless the backend is required to be changed for a specific example.
Thanks.
Thoughts?
Gavin.
-- -- Howard Chu Chief Architect, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/
-
Gavin Henry -
Howard Chu