Transforming signals to knowledge
The Intelligent Organization observes well. It transforms incoming signals into information and knowledge and responds adequately. This is only possible when the different processes within the Business Intelligence cycle are supported by a good architecture.
Link between processes and the BI applications
The architecture ensures that the processes run smoothly and that they are properly aligned. The Business Intelligence architecture forms the link between the processes we described earlier and the applications and Business Intelligence tools. Without a well thought out architecture, we cannot properly organize the processes of the Intelligent Organization and consequently, we cannot apply information and knowledge in the way we should.
All BI processes
The Business Intelligence architecture applies to all processes: from registering and collecting to distributing and responding. It should ensure coherence between financial and non-financial information, between internal and external information, between KPIs, viewpoints and processes, and between applications and tools.
An holistic approach
It is – among other things – this holistic approach that makes that Business Intelligence can deeply integrate in the business operations and that the organization can perform better persistently. It is therefore important to both develop and maintain a Business Intelligence architecture, which perfectly fits the organization and its business processes as well as the existing IT infrastructure and which is based on the desired indicators and viewpoints.
Prevent the system from instability
An architecture is necessary to ensure that both the Business Intelligence system and the Business Intelligence applications are (and remain) easily expandable, maintainable and flexible and that they do not cause major delays during expansion. The architecture should also prevent the system from instability. We should be able to easily add new indicators, reports, sources, instruments and data. After all, in today’s dynamic times, our information needs change much faster than before.
Definition of a BI architecture
Definition of (software) architecture (IEEE 1471-2000) “Architecture is the fundamental organization of a system embodied in its components, their relationships to each other, and to the environment, and the principles guiding its design and evolution.”
Designing a good architecture is an art in itself
A proper architecture can also ensure that information and knowledge are in fact reliable, for example through automatically checking the quality of data according to a fixed pattern and then – if possible – choose to either correct or reject the data. Designing a good architecture is an art in itself: thought and actions should go hand in hand. After all, we cannot simply think of everything on paper. We also need to go about things pragmatically in order to verify our assumptions in practice at an early stage.
The basic principles
We will look at the basic principles (successively) that rule the Business Intelligence architecture. We will also discuss the main pillars of the ideal architecture of the Intelligence Organization. These are:
- Data warehouses and ETL: collect, filter, cleanse, integrate and aggregate data in order to increase our chances to actually find information and knowledge.
- Portals and reflex BI: provide the possibility to directly apply and actually use information and knowledge in both the business processes and the systems supporting the business processes.
- Metadata: can be considered as the glue that holds all the Business Intelligence processes, applications and tools together allowing us to obtain an integral and reliable picture of business operations, at any given moment.
We speak of an ideal architecture
Metadata ensures that the ideal architecture can in fact support the Intelligent Organization from A to Z: from integral observation to automatic knowledge-driven responses. Although we speak of an ideal architecture, in practice a data warehouse is not always an option due to the costs involved.
Collection of cubes
A collection of information cubes may also suffice during the initial stage of developing an architecture. Although this is not ideal with regard to maintainability, response times and accuracy, what matters is that the management information is actually put into use by both managers and employees. Note that when we speak of data warehouses, these may also be considered as a coherent collection of information cubes.
Many viewpoints on BI architectures
The world of Business Intelligence houses many different viewpoints with regard to architecture and the methods of data modeling the data warehouse (Inmon, 1996; Kimball and Ross, 2002; Kelly, 1997; Van der Lek, 2003). Many of the differences in opinion concerning the architecture of data warehouses turn out to be theoretical though and unfortunately, there still seem to be organizations are guided (strongly) by one specific theory in this field. These organizations go after some guru without knowing exactly what is at stake.
Designed by the book but hardly useful
For that reason, we still regularly come across Business Intelligence systems and data warehouses that were designed by the book but hardly provide any meaningful reports. Worse even is the fact that some organizations seem to be proud of this – not so much of the reports, but of the fact that they did things by the book. Passionned Group covers the architecture from a practical point of view and provides basic answers to how and why. It also describes the purpose of the various elements of the architecture. In this way, it is up to you (the reader) to judge what is applicable for your specific situation and organization.