Example: Associating servers through referrals and basic replication
Following are the steps involved in distributing the namespace using referrals.
- Plan your namespace hierarchy.
country - US company - IBM, Lotus organizationalUnit - IBM Austin, IBM Endicott, IBM HQ
- Set up multiple servers, each containing portions of the namespace.
Following is a description of each server:
- Server A
- Perhaps just a server used to locate other servers in the US. With no other knowledge, clients can come here first to locate information for anyone in the US.
- Server B1
- A hub for all data pertaining to IBM® in the US. Holds all HQ information directly. Holds all knowledge (referrals) of where other IBM data resides.
- Server B2
- A replica of Server B1.
- Server C
- Holds all IBM Austin information.
- Server D
- Holds all IBM Endicott information.
- Server E
- Holds all Lotus® information.
- Set up referral entries to point to the descendants in other servers.
- Servers can also define one or more default referrals that point to
more knowledgeable
servers for anything that is not underneath them in the namespace.The default referrals go in the configuration file, not the backend.
Note: The default referral LDAP URLs do not include the DN portion. - Putting it all together.
Figure 4, Figure 5, and Figure 6 show these same six servers, showing the referral entries in the database including the default referrals which are used for superior references. Also included in Servers B1 and B2 are sample definitions for replication, setting up Server B2 as a replica of Server B1. This ensures that these two servers remain identical. Servers B1 and B2 are on the same system, but use different ports.