News

Looping Data Warehouses and Transaction Systems

• By Alan Radding
• April 4, 2005

Traditionally, data warehouses have had a one-way relationship with the transaction systems that supply the raw data. However, with the advent of SOA, the data warehouse and the transaction systems can be looped, each feeding the other. This is key for effective business intelligence and business analytics. In his report," Service Oriented Architecture Assimilates Data Warehousing," Lou Agosta, an analyst at Forrester Research, describes the feedback loop:

Third-generation data warehouses go beyond forecasting and decision support with their high volumes, large numbers of users, and complex workloads to close the loop from the data warehouse to the transactional systems. This means:

* Bidirectional process: This process is needed to optimize the upstream transactional processing based on the decision support provided by the data warehouse and forms the differentiating mark of integrated business intelligence. Today, active data warehousing and closed-loop business intelligence rely on customer design and implementation.

* Near-real-time functionality: SOA is used to transmit messages from the data warehouse back to the transactional system of record. These messages update the results of customer promotions recommended by the data warehouse or indicate the source of products specified in the demand planning data warehouse. When the data warehouse is used to optimize the transactional system, it is closed-loop, integrated business intelligence at its best.

* Operational efficiencies: Without SOA, the methods of closing the loop are custom-designed, one-off. They are effective-and have even been accomplished using COBOL and mainframes-but are not scalable from a project development and maintenance perspective. With the addition of SOA, closing the loop by means of near-real-time transactions from an application server, Web services or similar infrastructure enables efficiencies of operations.