Data Warehouse Guidelines: Building Data Warehouses

Patricia Seybold Group's Mitch Kramer gives seven things to keep in mind when considering a data warehouse.

Embarking on a data warehouse project is a daunting task. Many data warehouse projects are under funded, unfocused, end-users are not trained to access data effectively, or there are organizational issues that cause them to fail. In fact, a large number of data warehousing projects which fail, fail during the first year.

According to Mitch Kramer, consulting editor at Patricia Seybold Group, a strategic technologies, best practices, and business solutions consulting group based in Boston, there are many ways to make a data warehouse successful.

Here are a few of the areas to be aware of when creating and implementing a data warehouse:

1. Keep things focused.

“Try not to create a global solution.” Kramer suggests that a good practice is to “focus on what you need. A small data warehouse or data mart which addresses a single subject or that is focused on a single department is much more efficient than a large data warehouse. You will see measurable results much faster from a data mart than a data warehouse. A focused data mart will get funding and gain organizational consensus a lot easier, too.”

2. Don’t worry about integration, keep things small.

“Integration can be an issue, but it has always been a problem when organizations try to take a small filing system and integrate it into an organizational system. There are always coding problems of some sort.” Kramer then added, “global systems always tend to fold, so keep it small.”

3. Spend the extra money if you need help designing your system.

Kramer commented, “systems designing is the best place to spend the money on hiring consultants. They know the problems, and know how to deal with them. It is possible to design your own data warehouse system, but it is a lot less frustrating to hire out the design process.”

4. Keep things simple.

“Buy one single product from one vendor. This minimizes, or possibly eliminates any tool integration issues,” Kramer advised.

5. Be in tune with the users.

“Know your users,” Kramer warned. “If you are not careful, you will wind up giving the right users the wrong tools, and that only leads one place – frustration. Find out who your end-users are, and work backward to the operational data. This will tell you what tools your data warehouse needs.”

6. Consider your platforms.

Kramer said “there really are no right platforms out there. You can start with a UNIX system or NT. Keep in mind that the NT has a ceiling in terms of scaleability, but it works well with data marts, and most other small warehouses, just not global data warehouses.”

7. Think before you data mine.

“Data mining is a solution in search of a problem,” Kramer said. “Know what you want to find before you select the tool. Data mining software simply relieves some of the burden from the analyst.”

I need help with:

Got a Question?

Record it below. We’ll answer selected ones on our livestreams.

Affiliate Disclosure: This post may contain affiliate links, meaning we get a commission if you decide to purchase something using one of our links at no extra cost to you.