With the unwavering prominence of service-oriented architecture (SOA) there is an increasing interest in understanding what exactly it means for something to be considered "service-oriented." Thomas Erl recently completed a lengthy research project for SOA Systems Inc. into the origins of SOA and the current state of service-orientation among all primary SOA technology platforms. This body of work contributed to the mainstream SOA methodology developed by SOA Systems and was also documented in Thomas's new book, Service-Oriented Architecture: Concepts, Technology, and Design. We caught up with Thomas (a previous contributor to WSJ) to ask him to share some of the insights he gained from his work with SOA and service-orientation.

There's no need to mention that SOA has become a major focal point of the IT industry and a primary consideration on numerous corporate agendas. Nor is there a need to get into how SOA has been so heavily promoted that the term has already reached hall-of-fame status as one of the most recognized acronyms in IT history.

What is more important than the term itself is the impact its perceived meaning continues to have on how automation solutions are constructed. Its popularity to date is largely the result of vendors advertising SOA support or capability as part of their product lines. Because SOA has been so vendor-driven, its meaning has been somewhat divergent, skewed by proprietary technology that is still identified with common characteristics that transcend proprietary boundaries.

These common characteristics are critical to defining and understanding an abstract technology architecture classified as "service-oriented." Viewing SOA in abstract is what establishes an agnostic reference point from which proprietary implementations can be measured and, ultimately, unified.

Vendor-Oriented Service-Orientation
Vendors and other organizations in the SOA space have published numerous papers, blueprints, and even frameworks. Most such documents serve the dual purpose of educating readers about SOA while marketing related products or services. This is nothing new. Past variations of client-server and distributed architecture models have varied significantly in both technology and design, depending mostly on who and what was used to implement them.

However, because a core expectation of SOA is its ability to harmonize and streamline diverse technical environments, preserving an abstract viewpoint is required to achieving its potential. This is because SOA, when elevated to an enterprise level, can be used to establish an ecosystem in which an agnostic, overarching framework transcends proprietary environments and constraints.

How the components and elements within this framework are shaped and standardized is of critical importance. This underlines the need for a design paradigm that is sufficiently generic so that it can be applied to solutions regardless of implementation, while remaining in alignment with where powerhouse vendors and organizations are currently taking the technology that is fueling the service-oriented computing platform.

Service-Orientation and Object-Orientation
Design paradigms have played an important role in the evolution of technology and application architecture. The most widely recognized paradigm for distributed business automation has been object-orientation. The system-wide implementation technology for object-oriented solutions has traditionally been proprietary, where, despite the use of the agnostic principles of object-orientation, objects or components are designed to function and interact by using technology and protocols specific to a computing and/or vendor platform.

Service-orientation owes much of its existence to object-orientation. Like traditional multitiered architectures, SOA is based on a model wherein solution logic is distributed. As with object-orientation, concepts such as encapsulation, abstraction, and reusability are fundamental to the design of distributed units of automation logic (services) within SOA. Key differences in these approaches are focused on how these units relate to each other and the scope at which the respective paradigms can be applied.

See Figure 1

Service-Orientation and the Separation of Concerns
I have yet to find a better means of explaining service-orientation than to reach back to that fundamental software engineering theory known as the "separation of concerns." This theory essentially proposes that larger problems be decomposed into a series of individually identifiable problems or "concerns." The logic required to address or solve the larger problem can then also be broken down into individual units of logic that address specific concerns.

Past design paradigms and development platforms have applied this theory in different ways. Component-based and object-oriented designs, for example, provide specific approaches for the decomposition of concerns and the design of corresponding solution logic. Service-orientation establishes a new and distinct means of realizing a separation of concerns. As a design paradigm, it is an evolution of past approaches, augmented and extended in support of the overall goals and characteristics of SOA.

See Figure 2

Common Service-Orientation Principles
Service-orientation began with a modest scope - a basic set of principles centered on an architectural model focused primarily on distinguishing services as reusable and discoverable resources. However, technology architecture in support of service-orientation is making significant strides, and extending its reach into key realms of enterprise computing.

Expectations are being raised surrounding a new era of business automation composed of services as adaptive, shared software assets that promise to infuse an enterprise with organization-level agility, federated interoperability, and vendor independence. These expectations have placed demands on what a distributed automation solution classified as "service-oriented" should be capable of, expanding the breadth of the service-oriented paradigm and adding to and further shaping its principles.

So far, eight common and fundamental principles have been identified. Note that these are classified as "common" in that they represent a cross-section of the most widely accepted design approaches and best practices promoted and practiced by the organizations most responsible for realizing the contemporary SOA movement.

Here then are the common principles of service-orientation:

• services are loosely coupled
• services share a formal contract
• services abstract underlying logic
• services are composable
• services are reusable
• services are autonomous
• services are stateless
• services are discoverable