Mashed intelligence

As part of an effort to allow military planners to bring more data from varied sources together on one screen, Defense Intelligence Agency officials are taking a mashup approach. Their strategy resembles Google Maps. But DIA’s data and maps are classified, so instead of using Google’s services, the agency created its own system, known as Overwatch.

The application’s developers say they have gone further than any other Internet service by allowing users to easily toggle back and forth among different visualizations of the same data.

“Using these mashup capabilities, we can deliver any combination of information at any time to any user,” said Steven Willett, an information technology specialist at DIA who has been leading Overwatch’s development.

Mashup applications emphasize bringing together the elements of an application within a Web browser rather than relying on server-based integration and aggregation.

Mashups use technology based on JavaScript and Asynchronous JavaScript and Extensible Markup Language, a technique for updating Web pages that eliminates the need to reload the page. By making a client program responsible for aggregating data and displaying the results, a mashup can eliminate the need for a Web server to first collect the data.

However, an enterprise mashup must meet higher standards for security, governance, monitoring and availability. Overwatch does that with a layer of server software that regulates the entire process and provides the client-side code.

The system is based partly on JackBe’s Presto enterprise mashup platform, but DIA also created an independent, government-owned middleware layer written in Java and called the Agile Framework.

So far, the application has been deployed to fewer than 100 users at the Defense Intelligence Operations Coordination Center. Willett said a version of Overwatch will soon reach a broader audience as a publication vehicle for finished intelligence reports distributed through the Defense Department Intelligence Information System (DODIIS).

That expansion was made possible by the proliferation of DIA applications and databases available as Web services for publishing reports or querying applications.

In the past two years, the number of applications accessible as Web services has risen from about 15 percent to 50 percent, Willett said.

“I think this is actually driving the maturation of these services,” said Robert Ware, a DIA IT specialist assigned to the project. “Before, we would have to create a query for each individual. Now, we’re giving them the ability to create it themselves and share it through Overwatch.”

John Crupi, chief technology officer at JackBe, said trying to explain how the mashup approach is different is a common problem for his customers. One way the company has addressed that issue is by allowing its mashup software to run as a portlet within another vendor’s portal.

Willett said one of the mashup approach’s primary advantages is the way it incorporates applications and data from visualizations.

He has exploited that feature to change the way DIA adds data and applications to the system, cutting some of the overhead associated with getting them validated and approved. Typically, any application introduced on the DODIIS network must undergo a three- to four-month review and approval process, he said. By getting approval for a common user interface that can display visualizations for a variety of data sources, he has shortened that process. “We went from three months down to two weeks of testing,” Willett said.

Willett calls his approach the Dynamic Data Model View Architecture, a reference to the classic Model-View-Controller concept in which the data and logical model of an application is supposed to be separate from its user interface so that each can be modified independently of the other.

Willett said he aims to make the model/ view relationship dynamic, meaning that the user interface knows how to adapt to display the data that user queries summon.

“Nothing else out on the Internet does what we do, which is make visualization declarative, based on the data that comes to it,” he said.

The technique has applications that go beyond DIA’s uses, he said, adding that he hopes to drive further development of the technology as a government-owned software system that will be shared on an open-source basis among government agencies. Some, such as the Marine Corps, have already expressed interest, he said.