Departments
The View from Here
Brave new world
Shifting BI to the front lines is revolutionizing data warehousing.
by Stephen Brobst
We have entered a brave new world of data warehousing. No longer is a data warehouse considered an electronic filing cabinet whose primary purpose is satisfying reporting requests.
A best-practices data warehouse implementation has evolved from an isolated database platform providing a back-office reporting service to an integral component of the reference architecture for a real-time enterprise. The deployment of Active Enterprise Intelligence™ capabilities means that content from the data warehouse is used to deliver business intelligence (BI) to front-line decision makers in addition to traditional knowledge workers. Of course, the type of BI required and its delivery methods for the front lines will be radically different from those for traditional users of a data warehouse.
This evolution requires a rethinking of both technical and organizational boundaries within an enterprise. From a technical perspective, much more attention needs to be paid to integration between operational transaction processing systems and decisioning capabilities provided by the data warehouse. While the technologies used for online transaction processing (OLTP) and data warehousing are quite different, cooperation is needed between the solutions for delivering data into the data warehouse as well as providing decisioning services out of it. Organizational standards for enterprise application integration (EAI) and the use of service-oriented architecture (SOA) principles for building interoperable application services are essential. Integration of the data warehouse with enterprise standards such as lightweight directory access protocol (LDAP), in concert with platforms such as Microsoft’s Active Directory Services, is also advisable.
Organizationally, data warehouse architects need to work much more closely with enterprise application architects to ensure effective reuse of data within an organization. In the old style of construction, each new application required its own database to be designed and populated. Proper deployment of SOA principles will bring the processing to the reusable data rather than requiring new databases to be built. This approach has huge benefits in terms of total cost of ownership (TCO) and speed of delivery by avoiding duplication of data across multiple repositories. However, success with this more leveraged approach requires good governance. Specifically, this implies that enterprise architecture governance teams should include representation from a data warehouse solution architect. Lack of proper architecture governance leads to data anarchy. The result will be a proliferation of unnecessary—and expensive—data marts and operational data stores.
Proper integration of a data warehouse into the reference architecture provides huge leverage by encouraging pervasive BI capabilities whereby decisioning services are provided across an enterprise. A data warehouse kept in the closet for use by a small number of analytic rocket scientists will not yield the same magnitude of value as one that has been fully integrated for use by knowledge workers throughout an organization. The most advanced deployments of pervasive BI will go even further beyond traditional boundaries—enabling customers, suppliers, distributors and government agencies to use the data warehouse. At the heart of such new data warehouse architectures is the use of integration middleware to facilitate these capabilities.
Stephen Brobst is chief technology officer of Teradata.
Photography by Jeff Green