Intelligent Enterprise is revisiting what it calls the 34 subsystems of ETL -- which stands for "extract, transform and load," just in case you've been living in a IT vacuum the past 10 years.
ETL is a key data integration strategy. It's widely used, still very effective, and hence is unlikely to be usurped any time soon by SOA or any other Johnny-come-lately. So, the article on building and managing an ETL architecture is a great resource. It's written by Bob Becker, a member of Kimball Group -- hence the "Kimball University" tag before the title. If you have questions, Becker even gives his e-mail at the end of the article.
But what I found more interesting is actually in another article, a few pegs down on the home page, about how Pfizer Global Research and Development uses data integration middleware to eliminate ETL projects. To be frank, ETL was dragging everybody down -- IT, researchers, the finance people at Pfizer who are trying to make money back on all that research.
Pfizer Research does have data problems that go beyond your run-of-the-mill data challenges. The company must shift through hundreds of sources for anything relevant on a drug, including its effectiveness, possibly toxicity, patent information and so on. Portfolio managers need that information to make critical business decisions. Until now, they could only get it through IT, which used ETL to integrate the data from various sources.
To eliminate the need for ETL jobs and new warehouses, Pfizer created a portal environment using integration middleware from Composite Software. Now, developers can log into a portal and quickly access the data they need.
What once took three to four months for IT to accomplish can now be done in less than a week with the portal. And that's a tangible result for the business, which no longer has to wait months for IT to process and package critical business data.