On the Strategic Role of Legacy Systems in the Value
Proposition for Enterprise Interoperability
A position statement
by
Prof. Dr. Robert Meersman
Dr. Aggelos Liapis
VUB STARLab, Brussels, Belgium
Interoperability of enterprises is a strategic issue, caused —as well as enabled!— by the continuously growing ability of integration of new, legacy and evolving systems, in particular in the context of networked organizations.
To implement useful interoperability, (re-)establishing the correct meaning (semantics) of communicated business information is essential as well as crucial to success. For this, non-disruptive re-use of existing business data stored in “legacy” production information systems is an evident prerequisite. Note: by legacy systems we understand all existing computer-based “black box” information resources —this includes for instance, relational databases. Next, the scalability of any proposed semantic technological solution is another equally evident prerequisite. Yet on both counts current semantic technologies as researched and developed for the so-called semantic web may be found lacking. Yet in spite of that today, semantic technology is claimed about to become mainstream, as it rapidly becomes driven rather by enterprise interoperation needs (e.g. supply chains, value networks, …) and increasing availability of domain-specific content (ontologies) than by basic technology (e.g. OWL) providers.
To effectively evolve legacy systems in this fast-paced environment, managers require answers to the following types of methodological questions in order to align their systems with their partners:
• How can we systematically sort out all the issues with which we are confronted?
• How do we plan the non-disruptive evolution towards, and of, a large and complex interoperable system?
• What are the critical success factors for such an evolving system?
• How do we make sure the correct target groups are properly identified, involved and served?
• How can we determine if we are/stay on the right track?
• How do we implement/improve the interoperability of the system without adversely affecting our current local operations?
Because there is not yet an established methodology and discipline for systems and semantics evolution, organizations often focus on specific subsets of technical issues. In reality, however, many aspects of system evolution in the large are not strictly a matter of addressing technical problems. For example, there is a tendency to focus on a narrow set of technical issues and principles without considering the broader legacy systems (re-)engineering issues such as methodology, the strategic goals and objectives of the organization, and the business operations of the enterprise, such as an increasing role for collaborative approaches involving information resources created by the use of ICT in normal business practice. The latter in turn implies an
increasingly important role for ICT associated with and adapted to non-computer experts, usually derived with the help of natural language or other intuitive (graphical) interfaces.
As discussed elsewhere in the SIG report interoperability will in general be of different types and at different levels for any enterprise or enterprise information system. There can be physical (logistic) interoperability, data (syntactic) interoperability or business interoperability. As we are focusing on semantic interoperability it is important to expand on the role of ontologies since it they constitute the principal technological resource for enabling semantic interoperability between enterprises.
An enterprise framework should consist of seven elements that are building blocks for a successful system evolution or reengineering effort (Baker and Janiszewski, 2005). Each element has a critical set of technical and management issues that are essential for developing a comprehensive plan of action. The elements of the framework are
1. Organization 2. Project 3. Legacy system 4. Systems engineering 5. Software engineering 6. Technologies 7. Target system
Figure 1: A Framework for the Evolution of Legacy Systems (Baker and Janiszewski, 2005)
The framework that is being illustrated in Figure 1 may represent one independent organization. Our intention is to involve the methodology suggested through the framework displayed in Figure 1 into the interoperable framework displayed in Figure.2
In Figure 2 we describe that in order for two or more companies to evolve their systems into an interoperation mode they should not need to change their “local framework” for example their existing system- and software engineering technologies, their projects, their legacy system or even their target system. Systems will involve incorporation (e.g. wrapping) of state-of-the-art ontology management and organization of the collaborative aspects (e.g. workflows) of the companies that own them. Ontologies will be the backbone resources of computer mediated collaborative environments which will provide users with appropriate synchronous and asynchronous communication tools as well as recording mechanisms and advanced file repositories with automatic versioning support and group decision support systems (GDSS).
The development of a meta model compatible with existing legacy systems to foster interoperability along the product lifecycle traceability is considered to be essential. For instance, a well-known approach to such effect was pioneered by Winograd and Flores in their research in the area of computer mediated collaborative design environments and specifically applying speech-act theory in computer systems (Winograd and Flores, 1986). They were based on the study of the recurrent use of language acts inside interoperable organizations, with the intention of building a standard structure for action-oriented conversations, that means, conversations which have as goal to do something (a job, for example). They argued that inside an organization executives are part of a conversation network and the organization itself is concerned with maintaining this network. When a conversation does not result in a happy end, it is because there was a breakdown, a failure in the communication structure. Design then becomes concerned with breakdown anticipation.
Figure 3: A Winograd-Flores Structure for Action-Oriented Conversations based taxonomy of linguistic acts.
In Figure 3 lines represent actions and circles represent conversation’s states. A start the conversation making a request to B. In the state two, there are five possibilities for B: he (she) could accept, reject or negotiate a change on the initial conditions; Other two possibilities could be chosen by A: to cancel the request or to negotiate a change on the initial conditions. Following the normal course of the conversation, at the end (state five, seven, eight and nine), the actors will have a happy end. It is easy to notice that all the acts are linguistic acts and like a protocol, the acts have been conditioned by their antecessors. For example, any commitment act is followed by an assertive one (Winograd and Flores, 1986).
BAKER, C. D. & JANISZEWSKI, M. (2005) Essential Elements of EA Enterprise Architect, Fawcette Technical Publication.
WINOGRAD, T. & FLORES, F. (1986) Understanding computers and cognition, Norwood/NJ: Ablex.