Data Warehousing Key Terms

Making sense of data warehousing terminology

While large IT departments and companies have traditionally been the ones to take advantage of data warehousing, it's not out of the question for small and medium businesses. The high cost of the necessary software and hardware, coupled with the maintenance required, kept most small businesses out of the playing field.

If you're looking at the possibility of data warehousing options for your business for the first time, familiarize yourself with the data warehousing key terms first. If you're not IT savvy, you can get confused quickly. This guide addresses basic terms, such as data warehouse, data mart, queries, SOA, structured and unstructured data and meta data.

 

Data warehouse

Coined in 1990, the term data warehouse refers to a central collection (database) of data used as a repository and to support the business management decision-making process. The data collected is stable (you can add data to the data warehouse but can never remove it), subject oriented, integrated and identified with a particular time period.
Try: Refer to the article at LGI Systems Incorporated as you decide whether data warehousing is right for your company.

Data mart

A data mart is a repository of data (that may or may not come from a data warehouse) gathered from operational data and other sources that's designed to answer the specific demands of a single audience in terms of analysis, content, presentation and ease of use. A data mart is sometimes described as a scaled-down data warehouse for a single department or function.
Try: Expand your knowledge of data marts via the online Teradata Magazine article titled "Avoiding data mart traps."

Queries

Queries are concise memos with specific parameters for requesting information from a data warehouse. For example, you can write queries to retrieve quarterly sales values or comparisons of weekly, monthly, quarterly and annual sales values.
Try: Learn how to write your own queries from Penn Computing at the University of Pennsylvania Information Systems & Computing department.

SOA

SOA stands for service-oriented architecture. It's an enterprise-scale, conceptual IT architecture used to design applications and business processes where every computing unit of work equals one service. For example, a service can be determining inventory levels, ordering product or segmenting customers. SOA integrates your business as linked services and their outcomes.
Try: Put SOA into context by reading IBM's article titled "Service-oriented modeling and architecture. It paints a clear picture and references easy-to-understand diagrams and drawings.

Structured data vs. unstructured data

Structured data refers to data within fixed fields inside a record or file. The format of structured data remains the same; only the values change. Examples include spreadsheets, budgets, purchase orders, statistics, inventories and sales. In contrast, unstructured data comes in the form of free-form text such as emails or other word processing documents. Unstructured data also includes Bitmap objects such as image, video and audio files.
Try: Read the excerpt from "Tapping into unstructured data: Integrating unstructured data and structural analytics into business intelligence" for an easy-to-understand description of structured and unstructured data. For more information, check out the Bill Inmon's paper on the subject. 

Meta data

Simply and concisely put, the term meta data means data about other data. With regard to a data warehouse, meta data describe the warehouse's contents, data location, correlation to other data, sources and its relation to the company.
Try: Use the online document "Meta data: The key to data warehouse design" (a systems engineering approach) to gain an in-depth understanding of meta data.

Copyright © 2011 Business.com, Inc. All Rights Reserved.

---