While not OpenLDAP, I'm aware of testing done by Compaq with Active Directory of over 100 million Contact (similar to inetOrgPerson class) objects running on one server with a RAID array for storage. It was loaded with publicly available phone book data from all 50 states. California and New York were loaded twice to get over 100 million people. It was very cool, web based search method, very simple and very fast. I can't find the URL for it now, but it might still be on-line.
With the right storage engine, the X.500/LDAP data model scales very well. My apologies that I don't have any OpenLDAP experience to offer.
Charles Oppermann
Author, Microsoft Windows 2000 Active Directory Programming
http://www.coppersoftware.com/
-----Original Message-----
From: FOREST Laurent [mailto:laurent.forest@ca.sema.com]
Sent: Monday, March 05, 2001 10:32 AM
To: 'openldap-general@openldap.org'
Subject: Volume and Performance
Hello,
for a subscriber database, my company is thinking of using an LDAP server.
So I am looking at the OpenLDAP implementation.
The main question is obviously about volume and performance.
I retrieved some benchmark tools and benchmark results performed on NT
platforms from this list's archives. I also saw many questions but not a lot
of answers about the limits OpenLDAP can stand.
Nevertheless, I am doing one more attempt:
Has anyone heard of benchmarks performed on a UNIX platform for OpenLDAP
with a volume comprised between 1 and 10 millions entries?
Is it sensible or definitely non-sense (because it is well beyond its
limits) to further investigate an OpenLDAP solution for such a large
dictionary?
Thank you very much, and sorry if I have missed an existing answer to these
questions in the archives.
Laurent