Joe McKendrick hit on a great topic when he looked at the opportunities and the pitfalls of measuring SOA.
SOA has been taking a beating as of late because there have been major concerns about its ability to deliver ROI. But ROI is only possible if SOA-based efforts are properly measured.
Joe goes on to cite two recent reports around measuring SOA, but the core idea of placing metrics around SOAs and creating approaches for measurement of things such as ROI, has been a focus of mine for some time now.
The notion is that you can't figure out if SOA is working or not, unless you're willing to measure the effectiveness of the "to be" from the "as is." Thus, you need to create an approach to measurement prior to implementing SOA, and take action during the planning, implementation, and operational phases of an SOA.
There are two things to focus on.
First is the reuse aspect of SOA, which most of those talking about SOA measurement have done to death. This means how many services are reused within any number of applications and processes, and that number is pretty easy to determine if you have good SOA governance processes and technology in place.
Second, and often overlooked but more important, is the measured improvement (or lack of improvement) of agility. Agility is much more difficult to measure than reuse, thus people have a tendency to overlook it. However, it's typically way more valuable than reuse when you consider everything in the mix, including hard costs around the ability to quickly align IT with the business, the soft costs of customer satisfaction and employee moral, and overall ability for IT to finally work with business and not delay it.
Measuring both reuse and agility takes some thinking and some planning, but it's easy to establish the metrics and the approaches that work for your enterprise, and get a pretty good reading around the ROI of SOA. I suspect this will be more of a focus as we get better at SOA and need to determine the ROI to justify additional projects.