The purpose of replication is to make databases available to users in different locations or networks, on local PCs, or for failover. Eac copy of data is identical when replication happens, then updated databases will be out of step until they are replicated again. Replicated databases have the same replica Ids while copies of databases have different RIDs. Replicas can contain subsets of data, based on creation date, file size, modification date, document subset, deleted documents.
There are four types of replication
Replication in an HCL Notes Cluster is not scheduled but always happens when a database is changed, to ensure that data is consistent if the Domino server fails over in the cluster. This replication is also full, no subsets are allowed, to ensure all the data is consistent.
If you are running Notes as a local mail system, then by default Notes will not create a local replica of the mail database on your laptop. If you want a local replica you will need to manually create it like this.
Clustering provides both performance enhancements and resilience, by automatically replicating databases between servers in a Domino cluster. Performance is improved because the workload is balanced between the servers in the cluster. Resilience is improved by automatically switching users to good servers, if one of the servers goes down. Creation of a local replica of an NSF file on an HCL Notes client enables the user to fully use HCL Notes and databases while working off-line. The client synchronizes any changes when client and server next connect.
A CLUSTER consists of between 2 and 6 Domino servers, and is managed by the Cluster Manager.
A DOMAIN is all servers which share a common, public NAMES.NSF Address book.
n.b. A cluster is not necessarily the same as a Domain, and also Notes database names all take the pattern *.nsf
Every server in a Domino cluster has a Cluster Manager and a Cluster Database Directory (CDD). The Cluster Manager performs the switching and workload balancing functions described above, by exchanging messages called 'probes' with the other servers in the cluster. If the Cluster Manager needs to re-direct a user request, it looks in the CDD to find out which servers contain replicates of the required database. The CDD then, contains information on every database and replica in the cluster, including the pathing information needed to switch between servers.
If there is a CDD on every server, and they are pretty crucial to getting replication right, what keeps the CDDs replicated? The CDD Manager of course! This is a task which runs on every server. Other important system tasks are
The Cluster Replicator
It basically does just that. It intercepts any changes to databases on its server, queries the CDD to see if they should be replicated, then sends the replication probes to other servers in the cluster. Replication is determined at database level. Right click on a database, then select Replication then Setup. The icons on the right hand side describe the available replication options. Many of these are used by mobile users to conserve disk space. Replication can also be switched off completely both here, and also by an administrator using the Tools - Database - Replication options.
The Cluster Administrator
A housekeeper which Starts up the CDD Manager, The Cluster Replicator and other Administration processes. It also cleans up when a server is removed from a cluster, by stopping the clustering processes, and deleting the CDD.