For practical purposes, the chosen definition for Architecture (or Design) is: a theoretical proposal, components and relationships to achieve an objective, following a predetermined path over time, efforts and resources. Then, it will describe some combinations of components mentioned in Business Intelligence Concepts Article, from most basic design that fixes urgent necessities, to most accepted designs as best practices to resolve the most of information needs for all organization.
Do you figure out how to design...?
The conceptual design for Business Intelligence implementations could be presented as layered blueprint for data, processes, presentation and distribution components. Some of most common architectures are:
- One Tier data staging area: This is most basic set-up having a staging database accessed by end-user tools.
- Two Tier data staging area – datamarts: this is the evolution from previous model, having a best staging area (or several) feeding multiple specific-purpose datamarts.
- Three Tier staging – datawarehouse – datamarts: structured and specialized general staging area, feeding a general datawarehouse that transfers specific data to specialized datamarts.
Depending on several factors in every organization, it is possible to combine previous architectures in order to accomplish with successful Business Intelligence implementations dealing with specific constrains (mostly economic).
As initial layer for these infrastructure designs, is operational data or transactional data, external data sources, flat files (Excel and others), news and other disparate but important data types sources.
As final layer, we have the information front-end applications for diverse types of produced information. These applications could work together as a information portal or isolated as dashboards, regular reports, data analytics, ad-hoc queries and so on.