It takes 15 years, on average, to bring a new drug to the market, and each year of development may entail billions of dollars in research and testing. From an investment perspective, pharmaceutical companies want to make early, intelligent decisions on whether research on prospective drugs should continue so resources can be concentrated on the most promising projects. This dynamic led industry giant Pfizer to develop a portal-based data integration environment that has cut months out of the process of vetting potential new products.

Until last year, Pfizer Global Research and Development supported projects with custom IT development work, typically involving conventional ETL. Portfolio managers reviewing candidate compounds (or "drugs" in layman's terms) rely on the company's information gurus to scour hundreds of sources for potentially relevant information about effectiveness, toxicity, existing patents and scores of other attributes.

"We end up spending a lot of time working with data warehouses to identify the information we're looking for, but it's not always available for immediate consumption," explains project manager Dan Eng. "We have to go through a process of aggregating and summarizing data into a form that people can use."

Sources include databases containing everything from diagrams of compound molecular structures to test results from prior research to details on a compound's lineage through various stages of development. Also included might be business-side information such as the number of researchers involved in a project (noted in HR databases), the cost of personnel and other resources (noted in finance systems), and the number of "stage gates" each project has passed (as noted in project management databases).

"Between the administrative complexity and just getting at all the underlying sources, it used to take three to four months to go through the design, development, testing and deployment phases," says Eng.

Pfizer's approach changed in mid 2006 when it created a portal environment in which developers can quickly access multiple data sources through integration middleware from Composite Software. "Composite lets you bring together heterogeneous data under one logical view," says Eng. "Developers can tap into different sources and slice and dice the data without having to go through the laborious process of creating ETL jobs and new warehouses."

Pfizer's finance department had the Composite Information Server on hand for another project, and Eng says it took his team only a few weeks to learn the software and about two months to come up with the portal framework. That was fast work, but the biggest benefit by far of the abstracted approach has been the ongoing advantage of speedy data integration. Researchers and portfolio managers can access canned reports and a portlet-based query interface lets them select desired fields, filtering criteria and output formats including Excel and PDF. The bottom line is that what once took three to four months now takes less than a week, says Eng.

"The business can verify expectations much more quickly instead of waiting for us to design and develop," he says. "In three or four week's time, we may have gone through five or six different iterations and we could be a generation ahead of where we were before."