Enterprise Architecture on Ulitzer

Is SOA dead? This question keeps popping up every now and then within the IT community and though we all seem to agree that SOA is far from dead something has indeed radically changed: our expectations. Just a few years ago, perhaps months, we considered SOA to be the solution to all of our (IT) problems, promising to unleash the power of our business, and that just by thinking 'services'!

But judging by the decreasing number (still a lot anyhow) of events and focus around SOA it's quite obvious that SOA has been somehow a disappointment. The question is why? What did SOA promise that it did not deliver?

If we just focus on SOA’s technical aspects I do believe SOA has been immensely successful, I don't think there are a lot of companies where 'services' aren’t being considered or haven’t been implemented. Thanks to SOA we stress the value of standardisation and interoperability even more than in the past and even though we did build services before SOA was born, SOA brought a name and common acceptance to the discipline of ‘thinking services’. And it brought a whole lot of hype to the IT world which we most certainly needed.

But there’s more to SOA than implementing a technical platform based on services, once again we need to answer the question: what is SOA and what was it supposed to be?

The ones that did understand what SOA was all about did not succeed in communicating it to the big mass, ironically that probably contributed to it's success as it would have transformed SOA from something easy to understand and straight forward (on a theoretical level of course) to implement into something not a lot of companies fully have managed to achieve - the (at least partial) alignment between IT and business, ultimately an important component of a functioning enterprise architecture.

So, what would happen if we'd look at SOA from the EA perspective, its ability to permeate and contribute to your company's success? I am not so sure we'd like what we'd see.

Implementing the technical part of SOA (like building web services, deploying an ESB, ...) is a challenge by itself but that's far from the effort required to fully implement a SOA. In order to even start talking of services on an enterprise level you first need to create the correct organisational 'environment' and that is an enormous challenge.

Though companies and professionals embracing SOA did leverage the tools and know-how required (in the form of for example guidelines, whitepapers, expertise, etc.) to successfully implement a SOA, they evangelised the 'wrong' people, meaning often the IT people. That of course because they themselves mostly came from the IT-world and IT-departments were the ones they used to target. Imagine a sales-person from some SOA vendor going to the board of directors and trying to explain what SOA and services are and how 'thinking services' could benefit the company... and of course at the same time trying to sell an ESB (which is the sales-person's ultimate objective), it probably wouldn't be that successful.

So they went for the the bottom-up approach instead, sell first (the – technical - SOA concept and tools) and then explain why you are not going to build a SOA with just tools and technique, why governance, organisation, architecture are so vital to it and that without involving the entire organisation you will not succeed.

I'm not saying this approach is wrong, as a matter of fact I am not sure there is any other possible approach as once again, the alternative of going to the board and explaining the concepts of services, enterprise architecture etc. is not really that appealing... it's much easier to start off with an ESB and though it might not be the 'correct' way it's better to start the ‘wrong’ way than not starting at all.

That’s probably why SOA has widely been accepted and adopted by the technical community as a technical architecture, it’s the way it has been presented to it. And even though the concept of 'services' shouldn't necessarily be easier to understand for IT-people it seems that IT rushed to accept and adopt it, overlooking that it wasn't really an integral SOA they embraced rather the technical part of it and aspects such as organisational and political aspects were somehow partially or totally neglected. At least until IT realized that without fully implementing it SOA would never become more than exactly what they created, a technical architecture.

The ‘problem’ is that many companies never take the next step, they welcome SOA as a technical architecture and do not have the strength or the knowledge to implement it to it’s full extent into the organisation thus never unlocking it's true potential.

I guess you could say that SOA's inability to leverage what it promised is not really SOA's fault but lies on the not uncommon gap between IT and management which SOA itself was meant to 'bridge'. Catch-22.

So, what have we learned from all this? A lot. I think SOA helped us a long way on our path to creating a tighter cooperation between IT and the rest of the organisation and we have come much closer to creating an architecture that really ties together all the different parts of the company. It's like when OOP first was born and affected the way programmers thought, SOA made us think 'services', and even though different people thought of different kinds of services (making communicating pretty confusing) the important thing yet is that we introduced a common language that we can use to further develop our communication skills.

I believe that SOA will (or already has?) become part of our EA-toolkit as other architectures have done before it.