Jitterbit offers a fourth approach: integration by business analysts with no coding involved. Its Java client-based software allows business analysts to point to applications and data services or open the URLs of Web services to get a list of all methods, data types, and so on. They then drag and drop the components you want to tie together, using wizards to create a metadata repository of all those connections. With those relationships established, the Jitterbit environment provides the run-time engine to test the integration and then deploy it.

The use of metadata is crucial because that allows users to share integration work with others, whether business partners or someone down the hall with a similar need. Jitterbit CEO Sharam Sasson believes this will give smaller businesses the ability to more quickly integrate applications not just internally but also with business partners and customers.

CTO Ilan Sehayek acknowledges that the integration process requires upfront work to understand data types and methods, typically by the user delving into product and service documentation because Jitterbit can't know what the underlying service logic is. "But the next users will need to know a lot less," he says, given that the integration profile can store contextual information and even user notes, as well as sample data, to make the relationships clear. And Jitterbit hides the transport, security, and conversion layers from the user, handling the provisioning and execution automatically.

Jitterbit's technology is available in an open source form, though there's also a commercial form that includes EDI support. Further down the road, Sasson says the company hopes to deliver metadata-based workflow tools that allow the same type of abstraction both to let business analysts do the work and to let those workflows be shared easily with others.