Like the legendary comic, Rodney Dangerfield, 'SOA doesn't get any respect.' Burton Group have been leading an...
effort to rehabilitate SOA, most recently with a report on The Lazarus Effect: SOA returns. We spoke with Burton's Anne Thomas Manes about why the group believes SOA is not dead – and how IT depends on SOA. Ever objectively critical, Manes describes the six ways in which SOA efforts have failed.
If you really want to increase agility and reduce costs you must go to the source of the problem.
Anne Thomas Manes, Research Director with Burton Group,
SearchSOA: What's the basic argument here regarding SOA's undead status?
Manes: My colleagues at Gartner have produced a survey showing how many IT organizations are using SOA – 40% or more, they claim. My response is that I have no doubt those people are using SOA technology; things like enterprise service bus (ESB) and Web services, however, I don't believe many are really adopting the SOA paradigm. Therefore, they aren't really generating the benefits.
If you talk to anyone about why they are doing SOA, it is usually to increase agility and reduce costs. The problem is that just going out and using a different type of middleware isn't going to fundamentally change the applications and make them more agile or less expensive to maintain. If you really want to increase agility and reduce costs you must go to the source of the problem.
SearchSOA: And what do you see as the source of the problem?
Manes: Many organizations have invested a lot of money into their so-called SOA initiatives. They have bought all this middleware but they still have nothing to show for it. So, when they talk about SOA initiatives failing, they haven't recouped their investment, they have simply focused on technology. I have identified five reasons why SOA efforts have failed.
First, there are inappropriate expectations. Many vendors have said just use this ESB and you will achieve all these great benefits. But organizations underestimated the amount of effort and commitment required to achieve the substantial results that you can get from SOA. You will occasionally find organizations that have achieved the benefits but it requires a very serious commitment to make it happen.
Second, one of the surveys that Gartner ran a little over a year ago identified the fact that less than one percent of organizations have metrics that allow them to measure the value of SOA. So, even if they are generating benefits, they can't prove it.
Third, there is too much focus on technology. People were getting into SOA and proudly showing off their ESBs and such. But, they weren't looking at the architecture and design patterns and principles.
The counterpoint to that, my fourth point, is that there was far too little focus on architecture. In the end SOA is entirely about architecture. It doesn't matter what technology you use to design or implement services. What matters is the way you factor that capability into services and the dependencies you implement between different services, as well as the various components of an application.
Fifth is the issue of cultural resistance. One of the biggest issues is that while the funding model for IT is based on projects, often paid for by individual business units, those units want their things delivered as quickly as possible at the lowest possible cost, and there isn't much room in that model to look from an enterprise perspective at how to implement interfaces for things to be used in multiple situations.
Stop trying to sell SOA to business and just do it.
It is the project oriented model and the business unit-centric model that gets in the way of applying these SOA principles that are more global.
Finally, the sixth point is that there is insufficient governance. You have generally had every project running off on its own without a lot of organization and coordination. You hear horror stories about people deploying services and then you don't know who uses it and when it goes down, the production system goes down as well.
SearchSOA: Those are really interesting observations...So, what advice do you offer for people wrestling with SOA right now?
Manes: Stop trying to sell SOA to business and just do it. Essentially, you need to focus on supporting your business needs. So, become more business focused and understand what they need and then provide those things – and meanwhile, as you are doing that, you should apply SOA. You must learn SOA principles and design patterns. You must understand the paradigm and then apply it in every project you do and apply it at both the project level and the application portfolio level. You also need realistic goals based on the commitment your organization will permit you to make.