Public-sector organizations have a different way of life than commercial businesses, right? Having worked for both, I can say, yes, they tend to be driven by different objectives. But when it comes to SOA, things start to look awfully similar.
SOA still needs to prove its business mettle, even when there's no profit motive
I had the opportunity to attend last week's Open Group Enterprise Architecture Practitioner's conference in Chicago, and also participate in a panel discussion led by Dana Gardner and featuring Tony Baer, Eric Knorr, Andras Szakal, and David Cotteril.
Dana will be posting the panel discussion as part of his BriefingsDirect podcast series. One thing that was interesting to observe as part of the panel discussion -- as well as the conference in general -- are the similar challenges being seen by both commercial businesses and government agencies in moving to SOA.
One of the main points we touched on in the panel was the need for more convergence in the roles of enterprise architects, developers, and executives -- since SOA draws upon requirements for each. Do organizations need some sort of superhero type that can understand and bridge all these areas? Not necessarily, but stay tuned for the podcast for details on what skills are in demand.
We were all in agreement that David had the most enviable title: "Head of Innovation" for a UK government agency.
The OpenGroup conference focused on enterprise architecture issues in the government realm, but the message from many speakers was that SOA proponents within public sector bodies still had to make ROI cases for their projects, and be able to "sell" the initiatives to the "business" itself.
There were many terrific sessions offering a lot of detail. One that struck me was Tony Carrato's discussion of achieving ROI from SOA projects. He noted that there are a number of ways value can and should be tracked. Some of these metrics may include number of services developed in some period of time, number of services reused, and percent of reuse, "something or other about business flexibility," and decreased costs – usually in IT. (Carrato was fairly adamant, however, in noting that simply basing an SOA's value on the total number of services in production was not a good measurement.)
The important thing is that these results are measured -- which is sometimes hard to do, since SOA often delivers "soft" benefits -- such as business flexibility.
However, many organizations do not track the metrics to determine if SOA's benefits were realized. Carrato pointed to recent research that found that most organizations are "rigorous to very rigorous" when justifying a project, however, "comparatively few look back at the project, to determine if the stated value was received."
There are three categories of measurement, he explained: executive measurements (business results, pulse metrics, leading indicators); EA measurements (business & IT value, progress toward business & IT goals, compliance); and initiative and solution measurements for SOA specifically (quality, cycle time, cost, use satisfaction, business case realization).
Things that should be measured include cost savings across a portfolio of projects -- not a single project. In addition, organizations need to get into the habit of measuring the outcomes of projects -- and not just forecast the value of SOA before the project has started and leave it at that.
Open Group's main forte is Enterprise Architecture, and there's been a lot of speculation that SOA may be wholly absorbed into EA in the near future. An observation from Stephen Bennett also reflected on this evolution: that often, SOA may serve as a catalyst for engaging better EA practices. "I've seen some customers back into EA after starting an SOA project," he said.
Patty Donovan of the Open Group did a great job of summarizing a lot of session content from the event, particularly from the “SOA and the Boundaryless Enterprise” track. Here are some excerpts:
Pinaki Ghosh, lead architect specialist with The Dow Chemical Company, argued "that the main competitive advantage in information architecture comes down to a well constructed information footprint model within an EA framework, such as TOGAF, Zachman or DoDAF. During the transition from legacy architecture to SOA, however, one of the critical things most companies neglect is the preparation of an Enterprise Object Library. Such a library contains both business and IT objects categorized by international standards, unique artifact numbers and database identities. Pinaki’s presentation delved into The Dow Chemical Company’s use of a business object library to better align IT services with the business."
Michael Rollings, senior analyst with Burton Group, delved into Software-as-a-Service (SaaS), explaining "that although a SaaS solution may solve immediate issues, organizations blinded by its promise often ignore architectural challenges, integration issues, and other risks. After framing the business context for SaaS, Michael discussed several architectural principles which should be considered."
He was followed by Ross Button, vice president of technology leadership for CGI, who "examined the relationship between EA, SOA and Web 2.0. Button probed into the reality of the Internet’s influence on Enterprise IT and explained why EA initiatives must address modern Web 2.0 approaches, models and technology- both inside and outside the enterprise. Attendees got examples of several Enterprise 2.0 technology applications at CGI, spanning social networks, user generated content, mashups, SaaS, rich internet applications (RIAs) and SOA."