English abstract
Organizations are motivated to adopt Service-Oriented Architecture (SOA) as it provides many promised benefits such as, reusability, integration, business-IT alignment, flexibility, and agility. Several SOA Maturity Models (SOAMMs) have been proposed to evaluate the maturity of an organization's architecture and its capabilities to implement SOA; they define the SOA implementation process using different SOA maturity levels. Many Service Identification Methods (SIMs) have been proposed to simplify service identification phase in SOA lifecycle. However, a generic SIM that provides comprehensive guidance on service identification has not been developed yet. To successfully implement SOA, organizations need to select a suitable SIM with respect to SOAMMs, which should evaluate the maturity of methods used for implementing SOA. Due to the diversity of existing SIMs in terms of their features (e.g., input artifact, technique), it is difficult for organizations to select a suitable SIM to identify services based on the available input artifacts (e.g., business process, source code).
Many evaluation frameworks are available in the literature to compare the existing SIMs; however, an evaluation framework that addresses most aspects of the existing SIMs is still lacking. A few evaluation frameworks for SIMs have been proposed to guide organizations in selecting a suitable SIM according to their needs. These frameworks do not provide the desired guidance because they have not considered most aspects of the existing SIMs and also they have not used SOAMMs to guide the selection of SIMs based on the desired SOA maturity level. This study aims to: (1) propose a new selection approach for SIMs based on SOA maturity level to guide organizations in selecting a suitable SIM from the existing evaluation frameworks for SIMs to reach the desired maturity level in the SOA implementation process, (2) propose a new evaluation framework for SIMs that addresses most aspects of the existing SIMs to guide organizations in selecting the most suitable SIMs, and (3) propose a new approach for service selection based on Data Flow Diagram (DFD) to assist organizations speed up the process of migrating their legacy systems to SOA. The research methodology includes the following four activities. First, we conduct a literature review on service identification challenges in SOA. Second, we propose a new evaluation framework, which is mainly developed based on a review of the existing evaluation frameworks. Third, we propose a new approach for service selection based on DFD. Finally, we propose a new selection approach for SIMs based on SOA maturity level. This research produces significant results as follows. First, eight service identification challenges were identified from a literature review; also, the review discovered that service granularity needs further attention in the research community, compared to other service quality attributes. Second, a review of 23 evaluation frameworks for SIMs built the foundation for deriving a set of 16 criteria in the proposed evaluation framework, which addresses most aspects of the existing SIMs to guide organizations in selecting the most suitable SIMs. Third, a new approach for service selection was proposed that splits the services identified based on DFD into three service portfolios to select the best portfolio in terms of two service quality attributes (i.e., granularity, coupling). The evaluation results in two case studies (students' admission system, registration system) revealed that the first portfolio (i.e., all services identified from DFD level 1) had the best quality, as it achieves the two service design principles (i.e., high granularity, low coupling). The proposed service selection approach helps organizations in accelerating the process of migrating their legacy systems, modeled using DFD, to SOA by selecting high-quality services identified from DFD level 1. Finally, a new selection approach for SIMs was proposed that uses the desired SOA maturity level as guidance to help organizations in selecting a suitable SIM, for each level of SOA maturity, from the existing evaluation frameworks for SIMs. The proposed selection approach improves the maturity level of an organization in the SOA implementation process; also, it improves the applicability of the existing evaluation frameworks by providing the organization with a new way to select the methods. To fully utilize the proposed selection approach, we proposed a new evaluation framework that addresses most aspects of the existing SIMs to guide organizations in selecting a suitable SIM for each desired SOA maturity level.
