Recently, Burton Group declared SOA dead. Well, not really, if you read the report. The fact of the matter is that SOA architects are suffering through a bit of a hype hangover at the same time budgets are contracting. Thus, SOA is taking a beating. In some cases, enterprises are shutting down their SOA projects.
However, when you do a bit of digging, you’ll find SOAs that work. What they have in common are management support, the right approach, and defined and measurable value.
There are really two schools of thought around enterprise SOA. The first and most popular is that SOA is something you can purchase from a technology vendor; you simply bolt it onto your existing architecture and everything will be OK. However, enterprise architecture with systemic issues won’t get better by adding technology. Indeed, it gets worse. Thus, the fact there’s no quick fix or SOA-in-a-box has left those in charge of enterprise IT scratching their heads, and a lot of lost time and effort are being spent chasing unrealistic expectations.
The second and least popular school of thought is that in order to do SOA—which at its essence is good enterprise architecture—you need to approach SOA as something that requires a complete understanding of the problem domain, including data, services, and processes. Then, count on driving a core change to the architecture and selecting technology only after you have a detailed understanding of your existing domain, a plan to address the issues, and make the improvements. This requires the most work, but provides the most value.
Another issue has been the attempt to boil the ocean. SOA is about a strategy; your SOA should be done through a series of small, tactical projects where the value can be easily defined. Too many organizations take on huge, cumbersome problem domains, with outlined plans that span years, not months. These plans typically come to a halt when the company goes through any sort of business change.
A better approach is to define small, tactical projects around specific and measurable business problems. Thus, in months not years, you’re able to show the value of the project and SOA to those who sponsor the project. Of course, while doing these projects you need to adhere to an agreed upon vision and strategy that drives systemic and positive change to the enterprise architecture. However, doing that around many wins makes selling and protecting SOA a no-brainer.
So, where is SOA today? The architectural patterns of SOA pre-date the buzzword “SOA” and will continue long after the buzzword has fallen from grace. The trick is to define this architectural pattern as something that drives good architecture and provides the business with an architecture that can support the needs of the business, typically through architectural agility.
However, there are clear lessons we’ve learned as SOA has moved from a hype-driven buzzword to something more practical, and dare I say it, boring. These lessons include:
• SOA is something you do, not something you buy.
• SOA is best implemented through a series of small, tactical, high-value projects.
• SOA is about systemic change that provides architectural agility.
• SOA is driven by the business requirements, not technology.
• The A in SOA means architecture.
• SOA requires support from the top.
The next destination for SOA will be toward less hype and more results. Like other buzzwords from the past, we just stop talking and start doing. That’s a great thing, if you ask me.
