By way of background, let me take you way back to the proof-of-concept (POC), which is where the project I’m on had its genesis.
It was a competitive situation, where various vendors were given a set of requirements and a month to deliver an ESB-like solution that met them. Our team consisted of three people: Marty Wasznicky (Microsoft), Curt Peterson (Neudesic) and Todd Sussman (Neudesic). I was off doing other things at the time. Our team met all the objectives well ahead of the deadline, and then in fact proceeded to exceed expectations. At the end of the POC process, our team was the clear winner.
The basic capabilities of the POC were:
- Dynamic transformation (selecting and applying a map based on some external criteria, in our case the rules engine)
- Dynamic routing (contacting a UDDI directory [SOA Software] to get a SOAP endpoint URI)
- Integration with SalesForce.com
- Integration with AmberPoint
Transports involved were:
- SOAP (calling a SalesForce.com Web service)
- MQSeries
- File drop (as in all BizTalk demos and POCs :))
Although functionally the POC was very simple, it did prove out the various technologies, and showed that BizTalk was more than capable of playing a pivotal role in the client’s heterogeneous environment.
There were some interesting challenges with the SalesForce.com integration. Here’s what Curt has to say:
Integration with SalesForce.com's Web Services interface, at first glance, appeared pretty straightforward. Unfortunately, the WSDL that was produced by SalesForce included nested Schema references that confused the BizTalk Web Services adapter wizard; we ended up creating the artifacts manually in BizTalk. The good thing is, all of this could be done manually, the wizards didn't do anything "magic".
This is of course all fluid, and that statement was true in June 2005. Your results may vary now. Here an internal Neudesic Field Note that Curt wrote up that you may find of interest if you’re up against this.
This is the last background/intro post, after this we’ll start delving into architectural issues, and the timelines will converge with my somewhat hectic reality.