The question is no longer how best to automate business systems, but rather how to improve what's already been automated. We are nearly finished with the initial wave of IT adoption and most of what can be automated already has been. This means that the essential set of features and functions required to run a business already exists in some form. The trick is how to reuse and repurpose existing investments for additional value.
Everyone agrees that service oriented architecture (SOA) represents the opportunity for improving IT system flexibility and agility. However, the best way to realize the full value of SOA is to pursue a strategy of incremental adoption, rather than a kind of "big bang" or rip and replace approach. IT systems already in place have most of the features, functions, and capabilities needed by the IT systems of tomorrow - the key is to unlock that value through incremental improvements and turn IT system heterogeneity into an asset.
SOA is an approach, not a technology. The starting point is therefore not a product or vendor choice, but a decision to embark on a design and architectural process.
Advertisement
The process starts with the definition of services that the business needs and the data models and service contracts to support their transformation into executable software. Once the initial design is completed - mapping business services to software - existing systems can be inventoried and incrementally service-enabled to support the service design.
Some argue that the best approach to SOA is to start from scratch, creating completely new hardware and software infrastructure to support the new approach. Existing systems use older technologies not specifically designed for use in an SOA, and some additional investment is required to modify them. Removing existing IT systems and replacing them with completely new infrastructure promises to eliminate maintenance and support costs. However, deploying an entirely new infrastructure requires re-implementing existing business logic on the new systems while continuing to invest in parallel with the old, thus increasing overall cost and complexity. This approach also requires higher upfront investments for new software and hardware, while at the same time requiring a continued investment in keeping the existing systems running. And upfront cost increases are harder than ever to justify, especially when real business value may be slow to emerge, and given the high failure rates of new projects.
A "leave and layer" approach to SOA deployments benefits from incremental adoption. Instead of having to execute a plan that requires everything to be done at once, the leave and layer approach recognizes that significant investments in existing systems can be successfully reused. Most existing systems are stable and support mission-critical applications that have been in successful production for many years. Furthermore they typically run on software platforms that already contain the features and functions necessary for SOA.
The leave and layer approach allows existing systems' qualities of service (QoS) attributes to be leveraged in new applications. SOA infrastructure technology can be used to expose existing data and functionality as services, while preserving existing application reliability, security, and transactional capabilities. In fact, one of the most compelling benefits of the incremental leave and layer approach to SOA is that these QoS functions, along with existing communication protocols, transformation engines, and management tools that are already in place, don't need to be replaced. Even though existing enterprise systems typically were not designed to support an SOA, software products such as the enterprise service bus (ESB) can be used in conjunction with existing applications, preserving their QoS. SOA infrastructure products such as an ESB can very be very cost-efficient as they leverage what is already in place and provide a solid foundation for modern SOA based computing architectures.
The ability to deploy new systems incrementally enables the key value propositions of an SOA to be realized, avoiding the significant upfront costs of complete system replacement. With leave and layer, furthermore, you can find out as you go along how well you are doing with your SOA rather than waiting for some long anticipated day when you "turn the switch."
About the Author
In his role as chief technology officer at IONA, Eric Newcomer is responsible for directing and communicating IONA's technology roadmap and strategy.
Eric joined IONA in November 1999, after nearly 16 years at Digital/Compaq, where he held a variety of technical and management roles. He joined IONA as the company's transaction processing architect, and also served as IONA's Vice President of Engineering, Web Services Integration Products.
Eric leads IONA's participation in industry standardization activities, and has been involved in Web services from the beginning. He was a founding member of the XML Protocols Working Group at W3C, which produced SOAP 1.2. He serves as co-chair of the WS-TX Technical Committee and participates in the Service Component Architecture (SCA) initiative and Eclipse SOA Tools Project.
