11 Steps to Successful Data Warehousing

Only available on StudyMode
  • Download(s) : 62
  • Published : September 28, 2010
Open Document
Text Preview
11 Steps to Successful Data Warehousing
Mining your corporate data for valuable customer information can improve your business performance. But it's not as simple as it sounds. By Phillip Blackwood
There are 4 reader comments on this topic. Add yours!
More and more companies are using data warehousing as a strategy tool to help them win new customers, develop new products, and lower costs. Searching through mountains of data generated by corporate transaction systems can provide insights and highlight critical facts that can significantly improve business performance. Until recently, data warehousing has been an option mostly for big companies, but the reduced costs of warehousing technology make it practical -- often even a competitive requirement for -- smaller companies as well. Turnkey integrated analytical solutions are reducing the cost, time, and risk involved in data warehouse implementations. While access to the warehouse was previously limited to highly trained analytical specialists, corporate portals now make it possible to grant data access to hundreds or thousands of employees.

Following are some steps to consider in implementing your data warehousing solution. 以下是一些措施,考虑在实施数据仓库解决方案

1. Recognize that the job is probably harder than you expect.

Experts frequently report that 30-to-50 percent of the information in a typical database is missing or incorrect. That situation may not be noticeable -- or may even be acceptable -- in an operational system that focuses on swiftly很快地and accurately processing current transactions. But that percentage of error is totally unacceptable in a data warehousing system designed to sort through millions of historical records to identify trends or select potential customers for a new product. And, even when the data is correct, it may not be usable in a data warehouse environment.

For example, legacy system programmers often use shortcuts to save disk space or CPU cycles, such as using numbers instead of names of cities, that make the data meaningless in a generic environment. Another challenge is that database schema often change over the lifecycle of a project, yet few companies take the time to rebuild historical databases to account for those changes.

2. Understand the data in your existing systems.

The first step in any data warehousing project is to perform a detailed analysis of the status of all databases that will either contribute -- or potentially contribute -- to the data warehouse. An important part of understanding the existing data is determining interrelationships between various systems. Interrelationships must be maintained as the data is moved into the warehouse. In addition, the data warehouse implementation often involves making changes to database schema. You must have a clear understanding of data relationships among heterogeneous systems to determine -- in advance -- how any change may impact the system. Otherwise, it's possible for changes to create inconsistencies that ripple across the entire enterprise, creating enormous headaches.

3. Be sure to recognize equivalent entities.

One of the most important aspects of preparing for a data warehousing project is identifying equivalent entities and heterogeneous systems. The problem arises because the same essential piece of information may appear under different field names in different parts of the organization. For example, two different divisions may be servicing the same customer but have the name entered in a slightly different manner (such as AIG and American International Group). You can use a data transformation product that's capable of fuzzy matching to identify and correct this and similar problems.

More complicated issues arise where corporate entities take a conceptually different approach in the way they manage in-store data. This situation frequently occurs in cases of a merger or acquisition. The database schema of...
tracking img