Deploying replication in Db2 Warehouse

To begin using the replication feature in Db2® Warehouse on Cloud Pak for Data, you need to enable and activate the feature on the source and target databases, plan for key requirements and initial load of targets, and create replication sets.

You need two or more databases to get started with replication. Data changes at the source database are captured and transported through high-speed message queues to the target database. When you want to replicate in both directions, you need to set up replication in one direction at a time. Setting up replication for one direction is done from the source and automatically configures the source and the target for that direction. In this configuration, the two databases are both sources and targets.

Replication sets

A replication set specifies a set of tables for which transactions are to be replicated to a designated target.

Typically your replication set would include tables for which you want to maintain transactional consistency; for example, all of the tables that are used by an application. For a replication set, all dependent transactions such as transactions that modify the same row in a table are applied in source commit order serially. Independent transactions are applied in parallel on the target.

When you start a replication set, data changes for all tables in the set begin to be captured and sent to the target database. You can also add and remove individual tables in the replication set without interrupting replication for other tables in that set. When you add a table to a set, it can be automatically created, loaded onto the target system, and synchronized without the need to suspend transactions on that table at the source database.