Planning for replication

Before configuring replication, plan for network, user ID, and key requirementskey and referential integrity requirements, as well as for storage and initial load of targets.

Supported sources and targets
The following table shows the replication sources and targets that are supported by IBM Data Replication for Availability.
Network requirements
A private dedicated network for replication between the two Integrated Analytics System (IAS) appliances with a minimum 10 Gbps or fiber is preferred. To set up replication you will need the IP addresses and host names for the private network.
User ID requirements
The replication programs must run under a dedicated user ID with database administrator privileges. The apply program runs under this user ID, which also allows the capture program to identify the changes that are applied by the apply program and not replicate them back to the originating system. This user ID should not be used by any user applications.
Choosing the replication key
Replication requires a unique constraint on each table to uniquely identify each row across all databases where the row might be replicated. This is the replication key. The apply program specifies this key is specified in a WHERE clause for updates and deletes to ensure that the correct row is updated.
Considerations for RI constraints
When your tables have referential integrity (RI) constraints, include parent, child, and any other related tables in the same replication set. When adding tables to a replication set, add parent tables first and then the child tables.
Planning for storage
If your source database contains column-organized tables, a file system is used to stage supplemental log files that are generated by Db2 for each partition. You need to plan for disk space requirements on the source and target.
Replicating column-organized tables
For better replication performance, you can use column-organized tables instead of row-organized tables.
Workload and database considerations
Because replication uses log-capture/transaction-replay technology on sources and targets that are independent Db2 systems, it requires a few considerations to your workloads and databases.
Data type restrictions
Replication of spatial data types is not supported in IBM Data Replication for Availability, and some data types can be replicated only under certain circumstances.
Supported DDL operations
IBM Data Replication for Availability supports the following Data Definition Language (DDL) changes:

Parent topic: IBM Data Replication for Availability