ABSi's DW/BI reference architecture provides a template for Data Warehousing and Business Intelligence success
 

Closely aligned with best industry practices for Data Warehousing and Business Intelligence, is ABSi's Reference Architecture for system implementations. It is the foundation of our technical approach, and greatly follows the constructs of industry experts in integrating the various components of an effective Data Warehousing and Business Intelligence program. The architecture consists of the following key components:

• ETL (Extract, Transform & Load) – ETL components extract, transform, and load source data from enterprise transaction systems into the data warehouse and operational data stores::
  • Cleansing, integrating, and the intelligent management of source data, as a corporate asset is a critical task of any data warehouse.
  • Identifying, analyzing, modeling, extracting and transforming source data usually comprises 70-80% of the BI/DW project’s resources. The ETL process, therefore, needs to be tracked very closely during the project.

• Data Warehouse – The “centerpiece” of the architecture, it is defined as a subject-oriented, integrated, time-variant, non-volatile collection of data used to support the strategic decision -making process for the enterprise.
  • The data warehouse is the central point of data integration for the enterprise, and the “central source” of data for the Data Marts which “service” the downstream BI applications.
  • The warehouse also provides a common “view” of enterprise data (i.e., “one version of the truth”), usually via, what is called, a “star schema” design of organizing data along the lines of facts and dimensions. It provides the enterprise with a managed, historical data repository in one, centralized place. This is critical if the enterprise is going to do significant fact–based decision–making throughout many organizational components.

• Data Marts - Data Warehouse data is then used to support the specific analytical requirements and reporting needs of a given business unit, business function or application area. Data Marts serve this role.
  • Data Marts are “analytical databases” with highly multidimensional characteristics. “Facts” about the enterprise, which are “sliced and diced” by various “dimensions”.
  • These Data Marts “service” the BI applications, i.e., the reporting, query, forecasting and other analytical or decision support software that allows users access to the warehouse data.

• Meta databases - Provides the necessary details to promote data legibility, use, and administration. These Meta databases are described as  “data about data,” i.e., business activities, business rules, data transformation rules, and overall system and business process rules. They can be a rich source of corporate knowledge on the enterprises business processes and are vital in maintaining a corporate knowledge base.

• Operational Data Stores (ODS) – are defined as subject-oriented, integrated, current, volatile collection of data used to support the tactical decision-making process for the enterprise.
  • ODS’s are the central point of data integration for near real time oriented business management support. Their use is greatly proliferating due to the near real time need for analytical data.

• OLAP (Online Analytical Processing) enables users to have fast access to multidimensional data for their more complex analysis and reporting needs. OLAP applications support speed of thought data exploration, drill-down, slicing and dicing, etc.

• Modeling & Forecasting - Critical business strategies and what-ifs can be simulated using modeling & forecasting software. These tools can employ alternative business rules to test various business scenarios.

• Organizational Performance Management (PM)
  • Provides the enterprise key performance indicators (KPI’s) in a “scorecard” or “dashboard” framework for organizational performance measurements.
  • Measures lagging (“how did we do?”) and leading (“what’s going to happen?”) indicators. When linked with Activity-Based Models, PM applications can enable the enterprise to “simulate” activities that will change KPI’s so it can better accomplish its strategic business objectives.

• Statistical & Data Mining Applications
  • Applications for “discovering” meaningful new correlations, patterns and trends by sifting through large amounts of data, using pattern recognition technologies as well as statistical and mathematical techniques.

• BI Portals
  • Provide personalized, web-based access to key corporate data within the data warehouse environment. Portals manage the users’ data warehouse “experience” For example, log-on process with authentication and security, reporting and query navigation, self-maintenance mini libraries of individuals’ own reports and analysis.
  • BI Portals can provide access to multiple data repositories - both internal and external - and the ability to search both structured and unstructured information.

A sound architecture is a key success factor for an organization’s Data Warehousing and Business Intelligence program. ABSi’s approach is very architecture-based. We strive to make the components of the architecture modular - that is, we try to minimize lock-in to any software vendor and allow clients to evolve the architecture over time. So, as the data warehousing environment evolves and a change is required, our customers can make this change with minimum impact on the rest of the environment. Our reference architecture is not proprietary or an ABSi “invention.” It is an industry standard construct for Data Warehousing and Business Intelligence applications.