Monday, April 27, 2015

OpenDJ Replication Architecture

The way OpenDJ replicates data from one node to the other is slightly different from Sun/Orace Directory Server. OpenDJ utilizes Replication Servers for data replication as shown below:




I was reading OpenDJ Mailing List the other day and came across a great response from Chris Ridd:

The key difference in OpenDJ is that replication is not configured between pairs of servers like it often is in other products.

Instead, OpenDJ logically consists of a number of replication servers that are “fully meshed”, i.e. every replication server is connected to every other replication server. The replication servers hold the changelogs, but no LDAP data. OpenDJ also has directory servers that hold LDAP data that connect to replication servers in order to send and receive changes.

(You can configure an OpenDJ server to be a pure directory server (a DS) or a pure replication server (an RS) or both at the same time (DS+RS). The setup tool will create all of these combinations for you.)

When you use “dsreplication enable”, you are effectively adding a new node (--host1) to the existing fully meshed topology (--host2 can be any host in the topology). As Jean-Noël mentioned, the replication servers exchange views of the topology, so when you add a new replication server to the topology, it automatically becomes connected to every other replication server in the topology.


Good read!

1 comment:

  1. This comment has been removed by a blog administrator.

    ReplyDelete