By definition, a data group is a MIMIX construct used to control the logical replication of data between two nodes (systems). Clustering environments usually involve more than two nodes. In a clustering environment with three or more nodes, multiple data groups must be configured to ensure that data can flow between any nodes in the cluster. The total number of data groups needed to enable replication between all nodes in a cluster is known as the data group set.
In clusters with three or more nodes, at least one data group within the data group set is disabled at any given time. The data groups associated with the current primary node are enabled and data groups associated with only backup nodes or backup and replicate nodes are disabled to ensure that data from the primary node can be replicated to only the expected nodes. Disabled data groups are associated with backup nodes.
Only one user journal can be identified as the source of replication for a data group. Replicating from a second journal requires a second data group. Similarly, in a clustering environment, each source user journal is associated with a data group in a data group set. Replication from multiple source journals on a node requires multiple data group sets.
When starting or ending logical replication in a clustering environment, it may be necessary to invoke more than one command request to ensure that all of the selected processes for all data groups on a node have been addressed. This is because the three-part name of a data group definition only identifies systems, not system roles within the data group or node roles within the cluster. MIMIX can determine the role of a specified system within a data group but it cannot determine whether what you specify will select all processes for all data groups on a specific node. You can either determine the source system of each data group that includes the node you want and tailor your command requests accordingly, or you can adopt the practice of always invoking two requests which specify the data group definition as follows:
First request: DGDFN (*ALL *ALL node)
Second request: DGDFN (*ALL node *ALL)