Patterns: Information Aggregation and Data Integration with DB2 Information Integrator

276 Patterns: Information Aggregation and Data Integration with DB2 Information Integrator

– Changed data was captured in the relational database operational

systems using the appropriate Capture feature of DataPropagator™

(DPROPR on z/OS), DB2 UDB replication, and Oracle replication.

– The non-relational VSAM data changes were extracted from a journal

maintained by the rewards operational system.

Data mart details

As mentioned earlier, one or more data marts would be created on an on

demand basis for analyses and mining to plan targeted marketing campaigns

and/or alter the rating of a particular customer based on prior transaction

history. One such example may be creating a data mart of all customers in the

western United States having assets above $100,000 with a history of limited

transaction activity in their brokerage accounts in order to explore offering

possible incentives to them for becoming more active.

In such cases, the data mart data model would be a derivative of the data

warehouse, with the data warehouse being the only data source for

populating the data mart.

In our case we built a data mart with a data model identical to that of the data

warehouse (see Figure C-9 on page 258); the data mart DDL is also similar to

the data warehouse DDL described in Example C-6 on page 258.

Our scenario involved DFC exploring a new product offering (insurance

services) with customers in a high revenue generating region such as the

north eastern part of the US. We extracted customer data in this region (from

the data warehouse) having home owner demographics for further analysis

and mining in order to generate an appropriate targeted marketing campaign.

Figure C-19 on page 277 shows the simple model for loading the credit card

related data mart tables. It shows a single transform stage that connects to

the data source using the EXTRACT link, and the target table using the LOAD

link.

Note: Nicknames were created on the federated server for the tables

containing changed data, which were then defined as the data source for

Ascential DataStage.