The following is an excerpt from Agile Information Systems: Conceptualization, Construction, Management, by Kevin DeSouza. The featured chapter is written by John G. Mooney and Dale Ganley. It has been reprinted here with permission from Elsevier; copyright 2007. Download a free pdf of "Improving business agility with loose coupling and a Web services oriented architecture."
Web Services Oriented Architecture
A Web service can be defined as an independent and reusable component of functionality that can be invoked "over the Web" from any other component. The World Wide Web Consortium defines a Web service as "a software system identified by a URI [RFC 2396], whose public interfaces and bindings are defined and described using XML. Its definition can be discovered by other software systems. These systems may then interact with the Web service in a manner prescribed by its definition, using XML based messages conveyed by Internet protocols." By using Web services, application developers can produce modules that can be published once, yet invoked from anywhere across the Internet. This standardization of the connection methods combined with the ubiquity of the communications platform increases the potential for reusable modular functionality to a very wide audience. Web services-enabled applications can interact with other applications, or other Web services, to provide a level of connectivity that, up to this point, has only been a promise of the Internet.
The design and purposes of Web services are very similar to Electronic Data Interchange (EDI). EDI "refers to the computer-based exchange of standardized business-related information between buyer and supplier firms" (Hart and Saunders, 1997). The concept of standardized business-related information has been retained within Web services; what is new is the method of exchange. With EDI, participants are required to use a middleware platform to translate the data of each idiosyncratic application to a common form, then transport them from application to application, often translating again at the distant end. These middleware platforms are typically expensive configurations of hardware and software, and have resulted in closed, relatively small communities of EDI networks. Thus, the purported benefits of EDI like faster and more reliable data exchange between firms (Wang and Seidmann, 1995) have been constrained to only small groups of companies that can afford to implement the technology. In the n-tier application architecture perspective, wSOA provides the same functionality as the EDI middleware platform concept, without requiring expensive and proprietary systems to implement it. Essentially, the addition of Web services components moves the task of translating the data to a common form (and back) to the application, where the XML-based interface is incorporated. The transportation of the data is covered by the same Internet protocols that are cheaply and almost universally accessible to any firm.
wSOA is the integration of Web services components with other information processing and data storage systems, which may or may not be built with an eye toward inherent agility. This creates an IT environment that is highly modular and extremely malleable at the system level, despite any limitations of the systems that are integrated. The inclusion of business processes that take advantage of the wSOA creates the platform for an agile information system that can assist in meeting strategic goals in a dynamic business environment.
Loose coupling can be traced back to the concept of coordination among parts, as described in 1952 in Ashby's Design for a Brain (Ashby, 1952). Ashby described systems as being composed of distinct or independent subsystems acting in a coordinated fashion toward a purposeful goal. Mitchell and Zmud (1999) suggest that one of the primary features of a loosely coupled system is its ability to enact change, while limiting or delaying the effect on the interdependent parts. Loose coupling describes a particular state of connection between disparate elements. Orton and Weick (1990) assert that loose coupling "combine[s] the contradictory concepts of connection and autonomy." They further emphasize that the relevant dimensions of a loosely coupled system are distinctiveness and responsiveness. They extend earlier work (Weick, 1982; Perrow, 1984; Weick, 1985) to define a system with high lethargy and incrementalism (low responsiveness) and high indeterminacy (low distinctiveness) as loosely coupled. Beekun and Glick (2001) redefine loose coupling in terms of strength, directness, consistency, and dependence. Thus, it is clear that a loosely coupled system is interdependent to some degree, but that the components do not act responsively to each other. Loose coupling describes the functionality of loose linkages, which may be tightened to improve control but loosened to enhance cushioning from external jolts (Beekun and Ginn, 1993).
In the context of this chapter, we are concerned with the loose coupling between a firm's business and organization processes, and the loose coupling of the IT systems deployed to support and enable these processes. Two perspectives are valuable for analyzing this set of relationships: the process architecture perspective and the network organization perspective. The process architecture perspective examines how a process is decomposed and how the components interact to configure a system optimally for a given purpose. This emphasizes the direct impact of the loose coupling on the system that the architecture describes. Sanchez (1997) explains how, in the design of organizational processes, a modular approach with loosely coupled components helps to create the ability to reconfigure the system readily by enabling new functional process variations without excess disruption to the functions themselves. The network organization perspective focuses on the relationship that is being loosely coupled, which is implemented by an information system, and thus emphasizes the impact of the loosely coupled IT on the organization. Kerwood (1995) takes this perspective, discussing the new trend toward loosely coupled, cooperative interorganizational forms that are enabled by IT.
The Benefits of Loose Coupling for Strategic Agility
The outcomes of loose coupling vary strongly with the type of relationship it describes. We adapt the perspectives presented by Orton and Weick (1990) to the particular situation of loose coupling between IT-enabled processes. The following discussion represents a consolidation of findings from literature from both the process architecture and network organization perspectives.
Many theorists have asserted that a primary benefit of a loosely coupled system is to stabilize the system in the face of environmental uncertainty, and to prevent the spread of problems across subcomponents. This is derived from the condition of system modularity, which is directly enabled by loose coupling (Weick, 1985). This does not mean that the system doesn't change, rather that the system is more likely to adapt to change smoothly and at a measured pace. It is for this reason that Perrow (1984) suggested that loosely coupled systems are more desirable in high-risk systems such as nuclear power plants. Even in less critical environments, the ability to easily manage relationships between elements in a system, possible through loose coupling, can be a valuable strategic capability. In organizations facing uncertain environmental conditions, such as a competitive and rapidly changing market environment, the ability to either decouple or loosen their linkages with external elements to moderate the impact of negative events can be highly desirable (Weick, 1982; Perrow, 1984; Orton and Weick, 1990)
Perrow (1984) strongly advocates the adaptability of a loosely coupled system as a great advantage, especially in situations in which the system should be able to adjust to environmental changes. Mitchell and Zmud (1999) suggest that by promoting modularity and therefore the identity, uniqueness, and separateness of elements, loosely coupled strategies foster a greater number of mutations and novel solutions than tight coupling. Weick (1976; 1982) also asserts the diversity of responses that are enabled by loose coupling.
By enabling the linkages between elements to be broken or reconfigured easily, loose coupling can promote optimal efficiency of a system. Weick (1979; 1982) identified the reduction in coordination costs made possible by loose coupling. Furthermore, companies can retain control and support of the modular functions that are integral to their core competencies while outsourcing other activities to specialist organizations, thus increasing the overall efficiency of the organization's operations (Grabowski and Roberts, 1999). Supporting this idea, Ettlie and Reza (1992) find that loosely coupled integrating mechanisms up the value chain were "associated with internal capacity-related performance measures."
At the programmatic level, loose coupling is a correlate of nonredundancy (Burt, 1992). If a program module is built to allow loose coupling with other elements, future functional requirements are much more likely to take advantage of the existing loosely coupled module than build a new, redundant module. As Burt points out, strong ties that may start out as nonredundant are likely to become redundant over time; weak ties are less likely to become redundant. Since nonredundancy reduces the likelihood of errors and increases the ease of correction and modification to a module, loose coupling tends to be associated with programmatic efficiency of a system, especially over time and with increasing complexity. Additionally, Hansen (1999) asserts that maintaining strong ties are significantly more costly than weak ones. Thus, a loosely coupled system is likely to be cheaper to maintain.
According to the weak-tie theory originally advanced by Granovetter (1973), loose coupling is efficient for knowledge sharing because it provides access to novel information by bridging otherwise disconnected groups and individuals, while avoiding the pitfalls of redundant information seen in tight coupling. Thus, by stimulating knowledge sharing in an organization, loose coupling can lead to quicker search and transfer of relevant information and faster new product innovation (Hansen, 1999). In contrast, strong ties bind change much more tightly than weak ones, leading to system inertia (Hansen, 1999). Therefore, not only does loose coupling promote the ability to create new ideas, it allows the system to adapt to implement new ideas.
Orton and Weick (1990) state very simply that "looseness produces flexibility." Their logic follows from the observation that task uncertainty can be mitigated through loose structures and incremental adaptation paths, which is the argument that supports the advantages of persistence and buffering. Further, Weick (1988) points out that "loose coupling enables local knowledge to be applied when responding in a timely manner to local conditions. Small deviations can be sensed quickly and corrective actions quickly applied." Thus, loose coupling promotes flexibility by allowing a "diversity of response to unanticipated events" (Mitchell and Zmud, 1999).
Loose coupling can also enable more proactive adaptation of the network. It can expand the options available to support an instance of a linkage. Strong linkages tend to lock the relationship, which lessens any potential competition to support a side to the relationship. Loose coupling enables substitutes for a resource in the relationship to be viable options for switching. Therefore, increasing the independence of the elements in a relationship increases the strategic flexibility (Sanchez, 1997; Beekun and Glick, 2001). Kerwood (1995) makes a similar observation in an intraorganizational situation: the ability to customize systems or products. Systems that are tightly coupled are relatively inflexible and are more constrained to competing on superior economic benefits, such as larger economies of scale or lower costs (Kerwood, 1995).
Sanchez (1997) echoes this when he defines flexibility as the condition of having strategic options that are created through the combined effects of an organization's coordination flexibility in acquiring and using flexible resources. In other words, it is an organization's ability to respond effectively to various aspects of a changing competitive environment. Simply, he asserts that resource flexibility is greater when there is "a larger range of alternative uses for each resource; when the costs and difficulties of switching from one use of a resource to an alternative are lower; and when the time required to switch to an alternative is lower." He associates these conditions with loose coupling in business processes. He states that the modularity that results from loose coupling "greatly facilitates the creation and realization of strategic flexibility by an organization" (Sanchez, 1997).
Web Services and Loose Coupling
Web services offer an alternative approach to "enterprise integration." Instead of tight integration between applications through customized code or proprietary systems, wSOA enables loose coupling of existing systems, from both the process architecture perspective and the network organization perspective. At the process level, it either creates or replaces the middle tier(s) in data exchange hierarchies. It forces modular process architecture upon the system, facilitating access to processes of both coarse and fine granularity. It standardizes the relationship to a finite set of transactions that can be tested and optimized and available to applications across technical boundaries. Consequently, wSOA provides the loose coupling that is required in modular process architectures, and promotes the process-level efficiency benefits that accrue in such an environment.
At the organizational level, wSOA enables the seamless transfer of information across boundaries that are otherwise technically or financially impractical to bridge. It encourages the creation of linkages between otherwise disparate entities while still maintaining a high degree of autonomy between the entities. Where earlier solutions would have required heavily customized and usually proprietary systems such as EDI, wSOA greatly expands the potential for creating relationships between systems separated by strong and powerful boundaries. This is synonymous with loose coupling between organizations or organizational divisions.
- Read another chapter from this book: Agile information systems: Aligning business with IT
- Read other chapters from data management books in the Chapter Download Library
This was first published in November 2007