Original URL: https://www.theregister.com/2007/06/14/soa_comment/

SOA - dead or alive?

Can it deliver on its promises?

By Clive Longbottom, Quocirca

Posted in Software, 14th June 2007 10:23 GMT

Comment Service oriented architectures, or SOAs, have been positioned as the great leap forwards in how IT will finally come to be the major facilitator to the business.

However, uptake around the world remains mixed, and the levels of understanding that Quocirca finds in organisations as to what constitutes an SOA remains relatively low, especially within the business community.

Discussions with vendors in the SOA space show that they also find getting the SOA message across is difficult.

This has led many parts of the analyst and press community to start predicting the death of SOA, that it has been just another architectural move on the vendors' part to try and get businesses to spend yet more money on technology for technology's sake.

Whereas in some cases this may have an element of truth, those companies that have made progress with introducing true SOA have been seen to enjoy greater business flexibility, cost savings, and ease of technical management.

The problem is that for every successful SOA implementation, there seems to be 10 that are providing few of the desired benefits. Many of these failures are down to poor understanding of the underlying principles of SOA; in particular, in the use of "loose coupling" of services. Here, the various technical functions that make up an SOA are made available as individual items that can be called by other functions on an ad-hoc basis.

This needs a scalable registry of functions that can manage the negotiation and agreement of the calling service's needs and the called service's capabilities - and many organisations have found the concept difficult to understand, and harder to implement.

Many of these organisations have then implemented a more "tightly coupled" approach, with services being tied more directly to each other. This mitigates the need for the negotiation of need and capability, but also means that each new solution has to be made aware directly of what services it can utilise - and this means that functional reuse is minimised.

This is not the only problem, and another growing area of concern that Quocirca has become aware of is in how SOA is regarded by both the business and IT; whether it is seen as a strategic project, or a strategic aim across many projects.

As an example of the differences that Quocirca has seen in SOA understanding and adoption, let's take two, very broad-brush, views within Europe. SOA adoption in Scandinavia is fairly broad, whereas adoption in the UK remains relatively piecemeal. Both geographies are working from a relatively strong financial base, and are facing similar business issues - so why the difference in SOA progress?

Quocirca believes the main difference lies in how SOA is perceived both within the business and the IT department.

In Scandinavia, the business and IT tend to be very well aligned, and SOA is being viewed as an inherent part of new solutions. The business couches its needs in process terms, and IT looks at how these new process needs can be facilitated - from a combination of existing and new functions.

These new projects coming through are still being viewed as independent, stand-alone items, with integration into other existing parts of the infrastructure being carried out as necessary. This is the "tactically strategic" approach; each project is essentially tactical, in that it is run to solve a specific problem that the organisation is struggling with at a specific time. The viewpoint is strategic in that as each of these projects completes, a long-term view has been taken around SOA, and each new solution can automatically interact with other SOA-based solutions.

In the UK, we've seen the a different approach. Organisations have got caught up with the overall strategic message of SOA, and see it as being a forklift upgrade of the whole infrastructure. The business still couches its needs in process terms, but IT wants to create a corporate SOA environment, based around notions of elegance in development, code reuse and so on.

This approach constrains the options for the lines of business. Whereas a stand-alone project would cost an amount based on solving just the one problem, the first strategic SOA project within an organisation has to pay for all the underpinnings for the SOA, which may be an order of magnitude higher.

Without high-level buy in from the business itself as to how this new, strategic SOA project should be funded, the line of business is forced to go for a lower cost solution, which in many cases will be based around using standard off-the-shelf applications or through continuing existing methods of solution development.

This second approach therefore stops any real progression towards true SOA - existing practices are maintained, and the best intentions of the IT department to introduce the SOA concept just cannot take root.

Many of the vendors Quocirca has been talking to have now recognised this problem, and are beginning to review how they message SOA into those businesses where SOA has not been well received, or where SOA concepts are struggling to get off the ground. IBM, at its Impact 2007 user conference held in Orlando, detailed its new approach of messaging SOA concepts into the business as point solutions based around solving business issues.

BEA, at its latest analyst conference, demonstrated moves to approach the lines of business markets from a strongly business process viewpoint. TIBCO, through its experience with Staffware, has been using business process messages for some time. Even Oracle, and to a lesser extent SAP, are adapting their SOA messages to be more business focused.

Hopefully, this new messaging will make understanding at the IT and business level more complete. However, the main key for those companies looking towards SOA as a silver bullet for their business is to regard SOA as a journey, not as a single solution. By all means create a vision of where you want to end up on the journey, but don't let this vision cloud the short term view on how to deal with immediate business needs. Regard each new business issue firstly in isolation, looking at how you should solve this, utilising SOA approaches to this one problem. Then, look at how this fits with the longer term vision, and ensure that the functionality being introduced does not take you away from this vision.

As new projects come through, you will find that you already have pockets of functionality within other areas of the infrastructure; these can then be included in to the new project, making best usage of functional reuse. After a few projects, you will find that SOA has spread across the organisation, beginning to make the non-SOA areas the islands of capability, rather than the SOA areas.

This approach makes fast, tactical, solutions converge towards the longer-term strategic needs - and can make it that SOA can still provide on the promise we all hoped for.

Copyright © 2007, Quocirca