In the world of service-oriented architecture (SOA), service portfolio management is not yet a hot topic, but Forrester Research Inc. is working to raise awareness of the basic need to know what services a business has before it begins a project.
Service portfolio management is a key topic in at new 22-page report from Forrester analysts, "Defining SOA Service Life-Cycle Management" by Larry Fulton, with Randy Heffner, Catherine Salzinger, and Katie Smillie.
Service portfolio management is one of three sub-lifecycles that compose the overall SOA lifecycle in the Forrester view. Service consumption and service creation are the two that are most often talked about in SOA discussions, but Heffner, a Forrester vice president, argues that service portfolio management is the most important.
"Hopefully, you start off by using services, but you'll have to create services as well," he explained. "But those should both be done within the context of service portfolio management where we are trying to manage a long-term portfolio to make sure we have the right set of services, and that they are being used when, where and how they should be. If we're creating things, we're creating them in terms of a coherent portfolio of services and not just in the sense of 'here's a fun service to create.'"
Once an organization has developed a detailed portfolio of the services it has and the business domains where they operate, that will drive the other two sub-lifecycles, Heffner said. "Service portfolio management is the grease that makes the other two work," he explained. "In a sense, the other two should be subservient to portfolio management. If you're doing service portfolio management right in relation to the other two, then you change the model of how you accomplish reuse in a way that the industry does not really talk much about yet. The industry, when it talks about reuse, focuses dominantly on the word discovery."
With a focus only on discovery of services that can be consumed and creation of services that cannot be found, Heffner said, you end up with an ad hoc approach to SOA development. Architects and developers either go out and try to find a service to fit into the application or create a new service.
Working with the Forrest approach to the SOA lifecycle, architects and developers beginning work on a new application would know in advance what services they had available for reuse and what they would need to build.
"With service portfolio management you've got a structured set of major business services organized by business process domain," Heffner explained. "So when you start a new project, it's not let's go see what we can find. It's this project is working in this domain, we have a portfolio there, how should this portfolio come to bear on this project. Reuse is planned into the project that way."
Forrester views service portfolio management as a planning model for SOA development.
Working with this model, Heffner imagines architects saying: "We've got a portfolio. We're doing these major business units of work. We've got people who know and understand this domain. We can validate that we are using the service the right way.
As with all well-organized planning models, service portfolio management helps eliminate surprises of the negative kind.
As Heffner envisions it: "If we've been planning our portfolio well, then when we see a service that we don't yet have, it's not a surprise. This is something where we knew that we would need that service at some point because we've been thinking about it and understand our business. We know what the major business units of work are and, therefore, what our major services should be."
Like every other part of the SOA lifecycle, maintaining a service portfolio is an ongoing process. The Forrester report says that one of the key challenges in that process is the assessment of individual services to make sure they remain cost effective and are applicable to current needs.
The report suggests that the SOA team, in the best case scenario a center of excellence, go through the portfolio and ask questions such as:
- Is this service relevant to today's business objectives? "This is an assessment of the service's current relevance," the report states. "Past relevance is no longer important, for the simple reason that business objectives change."
- Does the service fit today's business model? "This is an assessment of whether the service's interface is consistent with current definitions," Forrester notes. "Put another way, does the interface reflect today's understanding of the business? An example would be customer data models, which tend to change as businesses evolve. Services consuming or exporting out-of-data representations are ripe for a migration strategy."
- Does this service meet its intended business need? "This is an assessment of whether the service meets its business requirements," according to the report. "As requirements shift over time, the service may require enhancements to remain valuable."