Skip to Content

In the last weeks and months many (partially populist) articles and blogs have been published carrying headlines such as “SOA is dead!” and “SOA is too cost intensive and complicated”! Those articles caused many rumors and confusion in companies and organizations whether it is worthwhile proceeding or even starting SOA related projects or not.

We think that is not the question of “if SOA at all or not” but rather of “how to start“. It is correct when claiming that the majority of SOA related projects have failed so far. But instead of questioning the overall methodology or even burying this revolutionary concept, it makes in our opinion more sense to evaluate a little bit more in detail the main reasons for those high failure rates.

Analyzing the reasons for failure typically three different categories can be identified. Whereas two of them mainly focus the mid and long term aspects such as a proper service definition and implementations and the right buy in of the upper management, we would like to take a closer look at the third category which has short term the highest importance.

Short term the biggest problems are often caused by the overall approach. Instead of trying to just start small, meaning to chose a manageable business pain point or business requirement for implementation, many customers immediately work on a long term strategy or roadmap, on detailed models for organization and governance.

For sure setting up an appropriate organizational structure and governance model within an organization is essential for successfully establishing a service-oriented architecture in the long run. But seriously which oil company would start building large refineries or factories next to a potential oil field by just believing that there could be oil without doing some test drilling first? Or if you want to climb a high mountain you do not already start drilling safety hooks into the ground before the mountain even is in sight.

What we want to say with that is that not all energy, budget, resources and time should be spent on planning a long term strategy. Companies in these days of high competition and in the middle of a world wide recession need quick solutions and fast process improvements with a “ROI in reach“. Only if the CIO of a company sees the value of an investment, he/she will be willing to invest more in it.

How to proceed?

Why not start with a small “test balloon” where there is a fair chance of a quick success and where not too much resources and time have to be invested? If the first scenario is chosen in that way that productized enterprise services are used, the danger of missing governance does not exist. Additionally involvement of experienced external resources in the first project might be helpful in some kind of “coaching approach”. This helps to learn and adopt fast the use of a service-oriented architecture and to avoid typical pitfalls causing higher costs and delays including frustration of involved people. Anyway expectations for a first small project should not be too high, because everybody should keep in mind that the use of SOA means new methodology and thus an increasing learning curve. And as always if something new is used or introduced there might be smaller problems lowering the ROI in the beginning.

In spite of smaller starting problems we believe that with a step by step approach the likeliness of success will be significantly higher leading to further interest and investment in a service-oriented architecture. Then as a second step there definitely is also the need for putting in place a long-term SOA strategy including the overall architecture, a supporting organizational structure and the right governance. Because as according to Confuzius: A journey of a thousand miles begins with a single step!

And in spite of a high failure rate in SOA implementation projects there anyway already have been several successful SOA implementations happened where corresponding showcases and success stories are available.

So why waiting and hesitating? “SOA is alive and kicking” and with the right approach and support the likeliness of success is also high. And experts located in SOA Competence Centers like ours for Retail and Wholesale are prepared to contribute our part to a successful project!

To report this post you need to login first.

1 Comment

You must be Logged on to comment or reply to a post.

  1. Vijay Vijayasankar
    I hear you – taking baby steps is a nicer and safer way to attempt SOA.

    But this is not always easy even if you start small. First, it is hard to get a nice enough use case with manageable scope. This is partly due to ignorance that causes people to overlook good candidates.

    Let us say I have a solution where I can either write some ABAP programs, and maybe some BAdI implementations to solve a small problem. In my desire to try out SOA – I try to tackle this problem using some out of the box enterprise services. And to prove that one way is better than the other – I do it both ways for comparison purposes. In most cases, the custom enhancement work is quicker to finish because the team on the ground already knows it and is good at it, whereas there is a learning curve with doing SOA. Quality of the solutions in terms of bugs etc are probably identical. SOA might need some additional investment in training, infrastructure etc too. At the end of the project – from tangible results perspective, it is hard to prove that SOA is superior in a small project. So then we have to start the “long term” TCO conversation again – and without proof points, the decision maker might just find it easy to extrapolate the results from the small project and decide “no SOA here for near future”.

    Success stories don’t get the news coverage that big failures get. So the decision makers might be swayed already by the time they see the results.

    Finally, over-marketing of SOA did its fair share of damage. I believe that we have to give it some time to let the hype die and let people figure out the value in a non-marketing scheme.

    My personal belief is that by 2010 or 2011, SOA will start making a real impact.

    I would also like to suggest to SAP that they should step up the effort to get more and more conulting companies to start doing SOA work in projects. The other thing I would suggest is to better educate the account reps from SAP on SOA, beyond marketing talk.

    (0) 

Leave a Reply