Finding the Best Metrics for SOA

When Ken Yerves looks at his IT dashboard today, he no longer sees the traditional metric for application availability. “The whole concept of an application goes away with SOA,” says Yerves, CIO and senior VP at JM Family Enterprises. Now he looks at business process and service availability.

Adds Tom Holmes, JM Family’s VP of technology operations, “Every business process is shown [on the dashboard], and we can see how long the process is taking to complete.”

The dashboard displays the number of minutes that a service has been meeting an agreed-upon service level. When any process falls below the target, the dial goes red and the number resets to zero and then keeps track of how long the service is affected. Once the issue is resolved, the dial turns to green and the timer resets to zero.

At H&R Block, when developers are asked to build a system that can take advantage of services, Neal Shaw, chief architect, estimates what it would have cost to build the system the old way and shares that information with CIO Marc West. West uses the comparison to communicate SOA’s cost-saving success to the rest of the enterprise. Conversely, Shaw sometimes has to advocate for spending more. “We might be spending 20 percent more now to build a system, but I can show payback year after year [derived] from the reusable components,” Shaw asserts.

When Patrick Moroney, former senior VP and CIO at $27 billion health insurer Health Care Service, planned to introduce SOA there, he projected setting up measures for service usage trends, response performance and value. One type of value he quantified was complexity reduction. For example, the company had four subscriber eligibility engines, but SOA would reduce that to a single service, allowing three of the four systems to be phased out.