I'm preparing to upgrade our OpenLDAP installation from 2.3.x to 2.4.21. I've been through chapter 13 (Schema Specification) of the admin guide and read the entries in the schema section of the FAQ.
With 2.4.x, there are several more schema files that are loaded by default, vs. what we are using in our 2.3.x installation (core, cosine, inetorgperson, misc, and one custom overlay).
Is there any appreciable benefit to not loading schema files that I'm confident we won't be using (java, dyngroup, corba, probably others)? I generally avoid "kitchen sink" software installation and prefer to pare down what's enabled in a software package to the bits we need, but I can't imagine skipping a few schema files is going to save a significant amount of resources on a modern system with 24 GiB RAM. Is there any other reason to skip schema files we won't be using?
Thanks,
Tim
Tim Mooney wrote:
Is there any appreciable benefit to not loading schema files that I'm confident we won't be using (java, dyngroup, corba, probably others)? I generally avoid "kitchen sink" software installation and prefer to pare down what's enabled in a software package to the bits we need, but I can't imagine skipping a few schema files is going to save a significant amount of resources on a modern system with 24 GiB RAM. Is there any other reason to skip schema files we won't be using?
If you're using schema-aware LDAP clients which load the whole subschema subentry you save bandwidth and client-side processing. My web2ldap caches the pre-parsed subschema subentry for a part of the DIT but the initial loading/parsing of really big subschema subentries (>500 kB) is significant on older systems.
Ciao, Michael.
openldap-software@openldap.org
-
Michael Ströder -
Tim Mooney