In-Depth

Words to live by

• By Alan R. Earls
• August 1, 2005

According to Forrester Research analyst Phil Russom, there are several competing terms that are either equivalent to on-demand data warehousing, or are closely related.

Teradata, for example, talks about active data warehousing, while other vendors talk about real-time data warehousing. However, Russom stresses, real time should be distinguished as more of a push technology, while on-demand is more of a pull technology. “On-demand (generally associated with IBM) is often, by its nature, more attuned to the real needs of the business rather than to simply what IT is able to accomplish with more technology,” he says.

By contrast, real time “hurls data at users whether they need or want it,” Russom says. Although on-demand can sometimes deliver results in real time, in some cases queries can still take minutes or even an hour or two to yield results. “But that’s still much better than overnight,” Russom adds.

“The real-time data warehouse is optimized for bulk loading of many gigabytes of data,” Russom says. By contrast, on-demand typically focuses on small pieces of information as they are created or changed.

Then there is an operational data store which an enterprise often needs in front of a data warehouse if it is to achieve real time. Many operational data stores are home grown, Russom says.

Enterprise information integration is another term that often gets lumped with on-demand data warehouse. Russom says IBM’s WebSphere Information Integrator is a leading example of that.

EII is sometimes used as a replacement for real time because it can go and get the data when it’s needed. “It’s not a replacement for a data warehouse, it is a supplement,” Russom says. But EII has limitations. “It can’t handle large volumes of data, Russom notes, and if you want to do true data analysis as
opposed to mere access, EII can’t help much.”

Mark Beyer from Gartner also offers some entries for the lexicon. For instance, he notes, ad hoc, reporting and OLAP often get mentioned in conjunction with on-demand. Ad hoc, he says, means an undefined use for data. OLAP means “I have a defined use but don’t know where the analysis will go, and reporting means I do have a defined use.”

Back to Feature: Real Time is the Right Time