SOA in Action Blog

Let me make one thing perfectly clear: I may be a Joe, but I'm not Joe the Plumber. I know how to connect a garden hose or unclog a drain, but that's the limit of my plumbing skills.

The other day, in the basement of my house, the hot water heater cracked open, and water came gushing out onto the floor.

So, I grabbed the documentation on the water heater and went down to my local MegaHomeStore. Mega's processes were great. The salesperson picked out a water heater that matched the size and dimensions of my current heater. He didn't seem to need to look at the booklet of the old heater. Once the salesperson entered the order for the new heater, the system generated an order slip that I took to the front register and paid. That transaction triggered a call to a plumbing contractor that was part of MegaHomeStore's partner network. An hour later, I received a call on my cell phone that the contractor had picked up the unit and would be over at my house within the hour. All smooth, fast, and like clockwork.

Great stuff, very efficient and helpful. But the only catch was that the unit I ordered was the wrong type for my setup, which required something called a "power vent."  The contractor spent some time trying to figure out what units were in stock that would fit into that space, then had to return the old unit.

So, because the incorrect (or no) data was not introduced at the beginning of the process (that I needed a water heater with a power vent), the processes were kicked off -- store time loading and unloading the unit, more than an hour of the contractor's time traveling to and from my house. If only there were a way, or process, to capture the right data right from the beginning.

This silly example (not so silly when you have to take cold showers) can be extended to represent the conundrum SOA and business process management run up against -- the processes and the applications are well tuned, very efficient -- but the bad data passing through the system makes it all fairly useless.

Ash Parikh, Informatica's resident real-time data integration and SOA expert, has been talking a lot about this problem, with data integration needs taking a backseat to the application side of SOA implementations.

On Tuesday, December 9th, Ash will be joining Madan Sheina, principal analyst within Ovum's Software Applications group, in a Webinar to discuss the urgency of data integration within emerging SOA and BPM environments. Beth Gold-Bernstein, my colleague at ebizQ, will moderate.

As Ash and Madan will explain, most SOA and BPM efforts have centered on dealing with the integration of application silos and automating business processes at the application layer. With data being the backbone of these applications and business processes, many enterprise architects end up struggling with a number of data-centric issues and equipped with the wrong tools for the job. For example, issues caused by the inability to access diverse and fragmented data, integrating enterprise data only to find out after-the-fact that the data is inaccurate and inconsistent, and dealing with enterprise information that is typically delivered at various latencies.