Is back-sql production ready ?

[Mike Bloom <nanogbloom@beanfield.com>]

Hi,

I've been using ldap for a few years now, but I've come to the 
realization that I have a lot of mysql databases lying around that 
should be integrated and unified. Its time to move to something more robust.

I'm using openldap-24 and am thrilled with the responsiveness and 
extensibility of the software, but I'm curious about using mysql as a 
back end instead of bdb.

I've been able get unixODBC up and running without any problems, but I'm 
stumbling a bit over the 10,000 foot view on metadata and how to 
generate metadata to insert a samba 3 object.

After skimming for a few days, I found this lovely overview 
(http://www.openldap.org/faq/data/cache/978.html) which didn't inspire a 
lot of confidence in how easy it would be to support abstract schemas 
that we might want to host on this system.


From this discussion and subsequent threads 
(http://www.openldap.org/lists/openldap-software/200511/msg00504.html) , 
I'm concerns about the practicality and that  I might have unrealistic 
expectations for being able to deploy 250k records on my mysql cluster 
to service a series of ldap servers over back-sql.

Does anyone run back-sql in production, and what kind of records on what 
scale of system are you using ?

I have high confidence that I want to use ldap, but I'd rather right 
provisioning software that talks to a mysql database than to talk to the 
ldap database. To make things more complicated, I suspect we will be 
moving to oracle in the next few years.

Does anyone have any suggestions on a 10000 foot view of metadata  ?

Thanks.