Use cases highlighting T-ETL

The following four use-case scenarios are designed to highlight the potential benefits that can be achieved when using WebSphere DataStage and WebSphere Federation Server to consolidate data. In each case, a data consolidation scenario using a WebSphere DataStage job is first presented, it is developed further by showing how WebSphere Federation Server can be used in conjunction with WebSphere DataStage to reduce both runtime and resource consumption. It is also shown how the original WebSphere DataStage job can be modified in order to take advantage of the capabilities of WebSphere Federation Server. The end of this section highlights the traits a WebSphere DataStage job should possess in order to benefit from this optimization.

Figure 2 illustrates the configuration used to test the use-case scenarios

Figure 2. Configuration used for use-case scenarios




















Depending upon the configuration of the job, the data is sourced from a number of different UNIX systems. Similarly, the target may be located on one or more UNIX systems. The DB2 UDB API WebSphere DataStage stage is used to access DB2 sources, targets, and the WebSphere Federation Server. All source and target databases are non-partitioned. IBM Information Server (which includes both WebSphere DataStage Enterprise Edition V8.0 and WebSphere Federation Server V9.0) is installed on a dual-CPU Windows Server 2003 machine. Each job within the four use-case scenarios is a parallel job designed to take full advantage of WebSphere DataStage's parallel processing capabilities. Since there are two CPUs on the WebSphere DataStage server, the degree of parallelism used was two.

The use cases refer to the following tables of a hypothetical parts delivery business. The tables are physically located on one or more source systems, depending on the scenario:

A CUSTOMER table with one row per distinct customer key. Each row contains (among other things) the name and account balance of this customer, as well as a designation of the market segment this customer is part of.

An ORDERS table with one row per distinct order key. Each row also contains the customer key that placed the order, the total value of the order, the date the order was placed, and a code describing its priority. There are typically several orders in the database for each customer, but some customers have no orders, or have not placed orders for a long time.

A LINEITEM table with one row for each item that is part of an order. Each row contains the order key of the order it is part of. Typically, an order contains several line items. Each line item row references a particular part key and includes the quantity ordered, the date the parts were shipped, and the shipping method used.

A STOCK table that links part keys and supplier keys and keeps track of the number of parts on hand at each supplier.
Before Next