directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Johan Danielsson <j...@pdc.kth.se>
Subject Re: "Different database methods in Heimdal"
Date Tue, 07 Sep 2004 07:10:28 GMT
"Noel J. Bergman" <noel@devtech.com> writes:

> It appears that you compared the performance of using an in-process
> DBM-class database to that achieved when using an LDAP server over a
> UNIX socket.

Yes.

> We are working on a Java implementation of LDAP and Kerberos. The
> latter is able to use LDAP, but also has access to an in-process
> interface to an embedded LDAP server. We'll be interested in seeing
> how our in-process performance our results match with yours.

Sounds interesting. I guess re-measuring the Heimdal+OpenLDAP numbers
now would give better values, there's been suggestions on how to
improve the performace too, but I'm afraid I don't know enough of LDAP
to remember how. :-)

> Is your performance test code posted with heimdal?

I think it's all there; kadmin/add_random_users creates a bunch of
"random" users, and kuser/generate_requests sends requests to the KDC
based on a set of usernames. Since the kadmin API is not very
standard, you will likely have to modify the former, but the latter
should work with any KDC.

/Johan

Mime
View raw message