Information in a box: Data Warehousing Can Help You Make Better Business Decisions

Target Audience: Manufacturing and Development Companies, M&D Industry, Data Warehousing Interest, Database Driven Companies, Business Process Improvements, Accounting Consulting Advice, Data Analysts, Emerging Business Trend Interest, Potential Database Purchasers

If your company is like most manufacturing firms, you probably rely on several databases to keep your operation running smoothly. This generally works fine, but what if copies of all your databases were maintained in a single “box,” available for immediate access from any location? Think of how much easier it would be to get the information you need to make important decisions.

Data warehousing gives you just that capability. With it, you (or a designated data analyst) can plumb your corporate records for even the minutest data from any of your databases.

Information, please!

Data warehouses can provide information to identify or improve a host of manufacturing considerations, including:

• Supply chain management,
• Sales forecasting,
• Detection and management of product defects,
• Worker productivity,
• Emerging business trends,
• Unprofitable product lines,
• Market demand, and
• Standard vs. actual product cost variances.

Data warehouses are subject-oriented, so that all the elements related to a certain event or topic are linked. In addition, they’re “nonvolatile,” meaning data is never overwritten or deleted. And they’re integrated, so they contain consistently formatted data from all your operations. Data warehouses are also time-related, with data changes recorded to allow tracking over time.

Handling the data

Before developing a data warehouse, you must identify what you want to measure, what data is required to measure it, where that data is currently stored and who will use the data warehouse to generate reports. The better you understand these business questions, and how to extract the answers, the more valuable the data warehouse will be.

Consider, too, how your data warehouse will extract data from the diverse software applications your various departments use. Somehow, the data warehouse will need to gather and integrate data from legacy systems that may have developed some unique quirks as they’ve been modified over the years.

Even if your legacy systems are compatible with your data warehouse, you need to make sure the data stored in each is decipherable to the data warehouse. If, for example, your sales department system records dates as Nov. 14, 2007, while the logistics format is 11/14/07, the difference must be resolved before the data warehouse imports this information.

This requires you to conduct a “data cleansing” using extract, transform and load (ETL) tools. ETL tools will extract information from databases and transform it to a uniform format before loading it into the data warehouse. In some cases, a database administrator can write scripts to perform these functions, but vendor-supplied ETL tools are easier to use for ongoing data cleansing.

Don’t get trashed

Before making the decision to buy a data warehouse program, you must spend a great deal of time planning. When it comes to data warehousing, the expression “garbage in, garbage out,” can’t be overemphasized. Taking care on the front end can help you avoid buyer’s remorse down the line.

Sidebar: The price of data storage

Data warehousing can significantly speed decision making and improve business performance. It can also become expensive.

Because data warehouses vary widely according to the extent of the data they encompass and the size of the company using them, they’re often priced by usable terabyte (TB). A terabyte is 1 trillion bytes, or 1,000 gigabytes. For some perspective, consider that the U.S. Library of Congress has more than 70 terabytes of data.

Pricing varies, too, but you can expect to pay at least $15,000 per terabyte for your data warehouse. Thus, a 10TB configuration may cost $150,000 or more.

One way to offset some of that cost is with open source data warehousing. Open source software costs nothing, because developers create it specifically to share openly with others, but you’ll need someone to build the data warehouse itself — a process that can take months. If you don’t have the specialized talent required on staff, you’ll have to hire it, and that can cost as much as $100,000.

Back to M&D Newsletters & Articles 

Back to Tax and Business Update Summaries

To contact Feeley & Driscoll, please click here or call us at 1 (800) 392-6192.