[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: back-bdb performance with SLAP_NVALUES



> -----Original Message-----
> From: owner-openldap-devel@OpenLDAP.org
> [mailto:owner-openldap-devel@OpenLDAP.org]On Behalf Of Kurt D. Zeilenga

> At 12:41 PM 4/1/2003, Jonghyuk Choi wrote:
> >Looks like we gained this from refactoring...

Excellent. The code size has shrunk a bit as well.

> I concur.  And as the test only involved simple
> normalization and small entry test set, it is
> reasonable to assume that the performance
> gain of SLAP_NVALUES+SLAP_NVALUES_ON_DISK will
> be even more significant when more complex and
> large entry sets (greater than what can be cached).
>
> I intent to make this the default and then, in a
> few days will run
>   unifdef -USLAP_NVALUES -USLAP_NVALUES_ON_DISK
> on HEAD.

Sounds good. There's a nvals memleak somewhere with just SLAP_NVALUES; no
apparent leaks with SLAP_NVALUES_ON_DISK. I'm not going to bother tracking it
down since we won't be using it.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support