Legacy isn’t planned — it just happens. For many public-sector organizations, the challenges of operating with legacy systems are all too familiar. Legacy environments are expensive and time-consuming to maintain; and brittle code makes it hard to modify one area of the system without impacting other areas unnecessarily. Making even simple changes to business processes can mean rework in multiple systems. Legacy applications are often single-purposed and reside in silos, making it difficult to share data and interoperate with other systems within the enterprise.
Successfully overcoming the challenges raised by legacy IT requires a “legacy renewal” effort. Legacy renewal leverages new application architectures to address the underlying problems of poor maintainability, poor scalability and lack of flexibility. When building an enterprise-class architecture capable of supporting multiple systems and lines of business, service-oriented architecture (SOA) is considered an industry-leading practice.
SOA allows organizations to incorporate commercial off-the-shelf (COTS) software into their overall solution, thereby providing the enterprise with best-of-class functionality, including, for example, business process management, business rules management, enterprise content management and security.
Exposing this functionality as a set of shared Web services allows these core capabilities to be shared across the enterprise, thereby increasing utilization and improving return on investment. SOA drives consistency across all systems. Whether changing a business rule, a workflow, or a security policy, changes need to be made once, and in only one place. What’s more, SOA allows non-technical users to make changes without programmer intervention, thereby allowing organizations to adapt to new business demands without having to compete for limited development resources.
The benefits of moving to a SOA-based solution are clear. Now is the time to upgrade. By failing to act, an organization not only faces the risk of technical obsolescence, but also a dwindling supply of people who are capable of supporting their existing environment. At the other extreme is the option of replacing everything all at once. But the “rip and replace” approach presents clear risks of major disruption to the business. For public-sector organizations in particular, this is likely to be a non-starter, as continuity is essential in the provision of many, if not all, public-sector services. A gradual approach to legacy renewal, based on service decomposition and SOA, offers an ideal balance between the two extremes.
Service decomposition and SOA can help an organization gradually transition from legacy to a modern enterprise architecture without incurring the risk and expense of the “rip and replace” option. Service decomposition essentially involves identifying candidate services for migration, and the order in which those services will be built. Newly developed services are deployed to the newly implemented SOA. Legacy functionality, that continues to provide utility, can be “wrapped” with a Web service layer and “plugged” into the new SOA. By breaking the dependency on the underlying architecture, an organization can utilize a phased approach to service migration, thereby considerably lowering the risk of business disruption.
Organizations can perform service identification by taking a top-down approach, a bottom-up approach or a hybrid of the two.
The top-down approach uses business strategy to identify the core competencies of the business and the business processes that are required to support them. It’s important to note that not all processes are candidates for automation. An ROI assessment governs the choice, so that only those that deliver the greatest value are automated. Once process identification and selection is complete, those processes are decomposed into their constituent individual tasks, which then become the service candidates.
The bottom-up assessment involves looking at the organization’s current enterprise architecture, i.e., its technical architecture (infrastructure), information architecture (data) and application architecture. By analyzing existing systems, current capabilities can be identified. Are there current legacy capabilities that work well and that continue to provide essential functionality, or is there legacy functionality that is “good enough” and that can be leveraged temporarily as part of a phased migration strategy? Are there recent purchases, such as a data warehouse or enterprise content management system, that provide core capabilities that can be reused? These capabilities are then decomposed and used to identify potential service candidates.
By combining a top-down approach with a bottom-up approach, a gap analysis can be conducted in terms of the services required and the capabilities provided. This is then used to create a service “road map” that catalogs the necessary services and provides a detailed plan for their implementation.
Public-sector organizations must select an appropriate approach to moving to a more flexible, agile and responsive platform. By enabling a phased approach to legacy renewal, service decomposition and SOA can help manage the risks that inevitably arise from such a transition.
Todd Wolff is enterprise SOA lead at Accenture Software for Health & Public Service.
This article was published previously in Policy & Practice, the journal of the American Public Human Service Association.