Jorge Santos wrote:
I'm trying to allow management of the sub-suffix by a different
group,
to increase the perfomance and because of scalability
The former can be done with ACIs, and AFIAK, you'll only get the latter
if you set up database replication rather than database links.
If you want to follow the replication route, two different
configurations spring immediately to mind. In the first, you set up
"dc=mg" (which is an unusual configuration; "dc" is the short name aka
"domaincomponent", which normally is used to describe a DNS name:
redhat.com == dc=redhat,dc=com) on the master server, and create "ou=bh"
underneath it. Create an administrative account in ou=People,dc=mg, and
give that account write access to "ou=bh,dc=mg". Then, set up the same
root on your second server, and establish replication of the dc=mg
suffix to that server. You can make the secondary server a read-only
consumer, or if you like, you can set them both up as multi-master
read-write servers, in which case they each need a replication agreement
to the other.
The other configuration that's possible is to set up each suffix on its
respective server, and then create a read-only replica of the other
server's suffix on the opposite. In this configuration, serverA would
have a read-write dc=mg, and a read-only ou=bh,dc=mg; serverB would have
a read-write ou=bh,dc=mg, and a read-only dc=mg. Searches would work
against either server, and writes would be redirected by referrals.
Clients that don't follow referrals would need to write to the correct
server explicitly.
So.. which way would you rather go?