SOA Governance
Real SOA governance starts with policy management at design-time. To embrace SOA successfully in daily practice, an organization must ensure that each project remains aligned with the requirements and best practices of the business. When building a network of large complex applications, it is imperative that organizations refine the initial high-level business requirements into more detailed technical specifications, while maintaining traceability between requirements and specifications, to prove conformance.

Traceability also facilitates impact analysis whenever the business requirements change or are considered for change. Tracking progress against requirements promotes agility and indicates to IT when it is safe to deploy updates.

Requirements definition and requirements management solutions help planners centralize requirements capture, refinement, and traceability. These solutions improve information visibility and collaboration, while prioritizing requirements and visualizing the results to determine the right requirements to implement. Business analysts can track and trace requirements throughout the development life cycle to ensure conformance to customer needs and compliance with relevant industry or government regulations.

Visualize the SOA
Model-driven SOA employs industry standard methods and languages to visually specify and manage the composite applications that comprise a SOA approach - introducing formality and propagating automation throughout the development of deployable applications and their technical services. Modeling allows business architects, solution architects, and developers to work at a higher level of abstraction while applying lifecycle-wide quality assurance techniques through graphical analysis and automated checking. The higher level of abstraction hides details and allows users to have role-specific views of the intended use cases and functionality, letting them focus on the solution that is needed from the application before worrying about how to build the application. Users can analyze and test the correctness and efficiency of the intended design by simulating it to verify compliance with requirements.

The modeling approach to SOA development lets users ensure quality and lower costs through continuous error checking and consistency testing at each stage of development, from architecture through final implementation. It maximizes the efficiency of distributed network-based applications by revealing both the components and their interfaces, while reusing and redeploying services through architectural reverse-engineering of existing software; updating legacy services via creation of a SOA interface; and improved orchestration among existing components. Perhaps most important, the model-based approach enables developers to simulate and test proposed changes to previously deployed applications before launching updates, saving users from costly downtime and exposure to untested application updates.

Formalizing Your Approach
The tighter the integration between enterprise architecture and the SOA application development environment, the better. In a best case approach, the organization should have seamless integration between these two domains, and even support associated technologies such as requirements management and change management. The goal should be to enable unprecedented organization-wide collaboration that helps internal teams understand and formulate business strategies and develop IT solutions that meet corporate goals.

This translates into an uninterrupted workflow from concept to deployment, with end-to-end traceability and accountability, ensuring that each stakeholder always has access to the right data and the right technology. Requirements management is a full participant in the process, with the business case driving requirements and requirements driving application development. (See the sidebar on requirements as part of the overall life cycle.)

For example, this workflow can be supported through seamless integration between modeling notations and products supporting enterprise architecture, business process analysis, SOA application development, and requirements management. Integrated solutions provide the conduit for each stakeholder to communicate and review new and updated business strategy and implementation concepts and changes in a way that all stakeholders can understand. Impact analysis can be conducted throughout the organization, allowing all participants to evaluate potential changes before final decisions are made and implemented.

SOA concepts, designs, and implementations can be described, analyzed, tested, and developed at all stages of abstraction, from high-level orchestration of business applications and services to the design and choreography of composite applications of technical services. Simulation and testing of the designs at each level of abstraction are important, with the design model having the ability to call and invoke actual Web services as part of the larger simulation.

As the deployable composite applications are designed, the SOA modeling environment enables the development teams to test for application completeness, with the test model forming the basis for specification and design of missing services, including whether they should be acquired from an outside source or developed further through detailing of the design model.

Adopting Model-Driven SOA
Service-oriented architecture (SOA) is receiving acclaim as a method to implement an organizational methodology based on composite application development. To be effective, the abstracted approach associated with SOA requires clear definition and synchronization at both the business and design levels.

A formal model-based approach to SOA, proven to bridge the gap between business and IT, enables organizations to properly implement and manage the SOA environment. SOA, then, becomes an efficient business language and an effective method of mapping and planning the complexity of the deployed applications. Modeling provides the means of explicitly relating high-level business objectives with the specific capabilities that each set of services can deliver, facilitating construction and maintenance of comprehensive applications that meet the business and design goals.

What does the future hold? As always, it's difficult to foresee how specific technologies will evolve over time. However, it is safe to assume that we will see a significant evolution of processes and best practices as organizations adapt to the power, agility, and flexibility of SOA and the responsibility for organization that comes with it. With its ability to visually represent the landscape of the enterprise, both business and technical, the model-driven approach to SOA ensures that operations keep pace with changes in the business environment.

References

• The Telelogic SOA solution
• Paper on Model-Driven SOA