DATA WAREHOUSING: A CATALYST FOR OPERATIONAL INFORMATION SYSTEM IMPROVEMENT:
PART II
by Kathy Long

3) Manage meta-data as an architecture component

The definition of architecture is "the art or science of designing or building structures". The architecture describes the how the distinct models developed to meet objectives form a cohesive framework. If we take this approach, a discipline is assumed where meta-data is not an afterthought, but an integral component of the warehouse environment. The meta-data architecture would include models defining meta-data data, meta-data processes, and the meta-data technology infrastructure. The information documented in the meta-data models is critical to the closed-loop IT architecture improvement process.

Our definition of meta-data is more than "data about data". The scope of meta-data includes business information meta-data, information resource management meta-data and access support meta-data. The business information meta-data are the business rules. Information resource management meta-data describes the information created to manage the warehouse environment. This would include extract and load statistics and data quality information captured during the data migration process. Access support meta-data is information that assists the end-user in their use of the warehouse, typically "how to" meta-data.

The source for business information meta-data is the logical data model of the warehouse business information. This model is used to design the dimensional models. This model is required regardless of whether the data warehouse is a data mart sourced directly from an application system, or a hub and spoke enterprise architecture with multiple marts sourced from a logically centralized atomic data store. The model documents atomic business rules in common terms. It documents detailed requirements and serves as an abstract, self-contained logical definition of the data separate from its physical implementation.

Data mapping, data transformation rules, and data scrubbing rules are also documented as meta-data. The capability to store this meta-data is usually incorporated into data migration tools, but if the extract and load process is home grown this meta-data could get buried in program logic. Since the data migration process may incorporate multiple tools, the rules governing the data migration process must be documented in a consolidated meta-model. Other views of the meta-data model at a physical level include staging area tables, and extract and load process management tables.

4) Publish meta-data

The business rules, data migration rules and captured data anomalies should be published to all warehouse subscribers. Warehouse subscribers should have information on the quality of data in the warehouse so that they can correctly interpret the results of their analysis. When data falls short of standards, that information should be used to promote the need for enhancements to the operational systems.

The meta-data delivery application should provide a feedback mechanism for analysts to communicate with the team on any topic. Analysts can provide feedback as data consumers on the usability and quality of information. The analysts can enter requests for new information, post questions that can be re-published as answered FAQS, and enter issues or problems they have encountered in their use of the warehouse.

5) Institute a change management process

The closed loop IT architecture assessment process won't become reality without establishing formal interface agreements and change management procedures between operational and warehouse teams. The change management process is the vehicle for the warehouse team to communicate data quality issues and identify improvement projects. The change management process should be incorporated into the systems development methodology.

Long term strategic plans as well as short-term and in-work systems development and technology infrastructure plans should be exchanged between dependent systems on an ongoing basis. If a change is required in any publisher or subscriber system, it should be formally managed through the interface agreements and change process. The procedures should specify the activities and deliverables through the change management process including change definition, impact analysis, resolution, specification, tracking and implementation. The cleansing rules and transformation rules that the warehouse team defines as meta-data should be used by the operational systems development teams to evaluate and plan required enhancements. Edits need to be placed as near as possible to the data creation point to correct data quality errors.

Summary:

This article has provided a brief snapshot of the ways that warehousing plays an integral role in the information architecture improvement process. The critical elements of a closed loop IT architecture improvement process are a data quality management process that includes stewardship, a strong meta-data architecture and formal change management procedures. A critical success factor is that the culture of the organization must support data quality objectives and reward achievements toward long term goals. Too frequently, data quality improvement is a concept that everyone buys into, but few are willing to invest in. Once these critical elements are established, the organization needs to identify the high risk areas and keep chipping away at them with incremental improvement projects. The benefit will be a well orchestrated and efficient information systems improvement process that meets business needs.

---
Part I : 10182

For more information, see http://www.spectrumtech.com