There has always been a debate among IT professionals about the best technical strategy for integration. Some argue that the implementation of an end-to-end enterprise system that provides tight integration of data and real-time visibility of performance is best. Others opt for best-of-breed applications that offer rich functionality but may require additional development work to integrate with other systems. Finally, bespoke systems provide the opportunity to create a tailored solution that is tightly integrated but with obvious increased risks to the investment.

The emergence of service-oriented architecture (SOA), which is the concept of consuming data from an isolated system as a service and integrating with other isolated systems via messages, has further fuelled the debate.

Its proponents argue that SOA provides an easy way of integrating disparate systems, extends the types of information that can be integrated ­ documents and emails, for example ­ and reduces the reliance on one specific vendor.

SOA can be used in most organisations to add value. If you adopt SOA on top of a core end-to-end business system, it provides the ability to further automate the business processes and integrate additional information held outside the system ­ for example, linking documents and emails to objects in the system.

In fact, many software vendors have recognised this and are now offering SOA-enabled versions of their systems. Even bespoke applications can benefit from being SOA enabled, providing a mechanism for integrating and sharing information inside and outside the business.

SOA is without doubt a useful technology but there are risks, in particular the proliferation of metadata and an increase in maintenance and support of components. An unplanned and ill-conceived implementation will result in poor returns and a future legacy nightmare. But if used wisely, SOA can deliver real benefits.

Alistair McLeod is a BCS contributor