Tuesday, August 01, 2006

Data Syndication

We are in the process of migrating our existing clients onto our new version of CXP, the Customer Experience Platform. One of our clients has discussed with us the possibility of integrating their inventory data into the platform so they can build rules based on inventory levels. The term "integrate" in the past has meant that a lot of time and energy was going to be expended creating a tightly-coupled back-end data integration link. In recent years we might have said, "oh, sure, we can do that, just expose a web service and give us a WSDL file and we will begin testing the integration." Of course, a web services deployment is more complicated than just publishing a WSDL. It takes a lot of resources, both developer and infrastructure.

What to do about this? How do we achieve the long-promised benefits of internet-enabled applications without all the bloat? One of the possibilities that we are exploring is integrating data via data syndication. Similar to RSS, this method involves the owner of one system publishing selections of their data in an XML file. The second system then consumes that data and uses it in its functions. The beautiful thing about this model is that the data gets consumed on a web client or a browser versus integrating the data into the back end of a system. It keeps it lightweight for both systems. It makes the system truly loosely coupled. Of course, this method of integration is more appropriate for systems that do not have high dependencies on each other, and it is a one-way thing, but there are many many business applications for this type of integration, so why not run with it?

We have not yet built this type of integration yet, but I imagine we will in the next few months. I will blog it when we do.