IRS embraces service-oriented architecture

The Internal Revenue Service is looking for ways to modernize its online customer service without first going through costly back-end upgrades.

A service-oriented architecture (SOA) approach is allowing the agency to take data locked away in existing systems and connect it with Web applications, said Bob Albicker, the IRS’ deputy associate chief information officer for systems integration.

“You can take advantage of the SOA capabilities to, in effect, make it seem like you’re working with modernized data, when in fact, you really aren’t,” Albicker said.

SOA is a methodology that seeks to break down complicated business transactions into individual components. Once a complex task, such as “pay taxes,” is divided into components, or services, it is possible to reuse those services for other tasks. The ability to transmit data from one software application to another is crucial to this approach. In the past, data has often been tied up in proprietary systems and hence not reusable.

The agency has already been using point-to-point connections to shuttle data between existing and modernized applications, Albicker said. For example, on IRS.gov, taxpayers can request to see the status of their tax refunds, a function that requires retrieving data from existing systems.

But “that’s a tremendous number of interfaces to keep up in the air, and every time tax law changes and your return format changes, you have to go out and reprogram every one of those interfaces,” Albicker said.

Earlier this fall, the IRS selected webMethods’ Fabric integration suite to begin systemizing connections between existing and updated applications. The goal is to standardize the interfaces so that no matter what an application requests of a particular data set, there’s a single route to retrieving the information. That way, if tax law changes, “then all we have to do is basically write that change and test that change once through the integration broker,” Albicker said.

An SOA approach won’t negate the long-term need for modernizing systems, he said. “There’s only so long that you can continue to find and train people in assembly language code,” Albicker said, referring to the 1960s computer language that powers the IRS’ Master File of taxpayer data.