This article originally appeared on the BeyeNETWORK.
What is architecture? Architecture is the combination of the science and the art of designing and constructing physical structures. A wider definition often includes the design of the total built environment, from the macro level of the physical structure itself to the micro level of architectural or construction details.
An information system (IS) is the system of persons, data and activities that process the data and information in any organization, including manual and automated processes. A data warehouse is a subset of the information systems, whose purpose is to collect data from disparate sources and present it as integrated solutions to customers.
Data warehousing architecture is a complex subject. It is not a simple database on a server with a data model and processes to load and query data. Rather, it is the foundational layer for the business intelligence initiatives in the organization. A data warehouse is a program that will enable multiple projects. Its architecture and the blueprint that will drive its construction are critical to the success or the failure of the program and its projects.
Data warehouse architecture can be classified into the following areas:
- Business Architecture
- Technology Architecture
o Hardware Architecture
o Software Architecture
o Database Architecture
o Security Architecture
o IT Governance Architecture
- Data Architecture
o Data Integration Architecture
o Data Movement Architecture
o Metadata Architecture
o Master Data Architecture
o Data Governance Architecture
- Business Intelligence Architecture
o Data Visualization Architecture
o Data Querying Architecture
o Data Analysis Architecture
o BI Governance Architecture
As we list these major components, there are several interesting approaches to integrating the major components at the micro level in a given data warehouse. This brings us to another major component for a data warehouse program – methodology. While architecture describes what we need to build and lists the components, methodology represents the delivery mechanism of how to build the data warehouse and deliver the same.
Architecture and methodology need to work together for the overall success of any data warehouse program. While on this subject, architecture and methodology describe the technology and process legs of a three-legged stool. The other leg of this stool is the people or the data warehouse team.
There are several architectures available to choose for a data warehouse implementation. How do you select the right architecture for your organization? An easy technique will be to assess the components listed above in a weighted scorecard against the architectures and select the top two as your ideal choices.
As shown in this simplified approach, the four steps to selecting and implementing architectures can be easily confined to the process of defining the needs, assessing the choices, designing a solution on one or more selections and doing an easy prototype. This approach will definitely help you in the entire program, though you might choose to implement instead of prototype.
As we conclude this article, the fundamental goal of a data warehouse architecture is to present the blueprint and a road map to build a complex data processing and integrating infrastructure, while enabling the delivery process to control the build and deploy in a selected methodology framework.
My next article will discuss the different architectures across the multiple components by subject areas.