Change impact analysis model-based framework for service provisioning in a grid environment
Grid-based Utility Infrastructure for Small, Medium and Micro Enterprise (SMME) Enabling Technology (GUISET) is an architecture in which distributed applications must interact across platforms via the concept of services. The building block in a GUISET application are business process services which are expected to undergo maintenance just like any other software components. The basic operation of software evolution is change, thus making change inevitable in software development. Throughout the entire lifecycle of a software system, from conception to retirement, things happen that require the system to be changed. Changes are required to fix faults, improve or update products and services. Under these conditions, it is crucial to have full control over and knowledge about what changes mean to the system and its environment. Otherwise, changes might lead to deterioration and a multitude of other problems. The effectiveness and efficiency with which a company can predict or control these changes could have a significant impact on its competitiveness. In a complex product, where the constituent parts and systems are closely dependent, changes to one item of a system are highly likely to result in a change to another item, which in turn can propagate further. The activity of assessing the implications of realizing a change is termed change impact analysis. One glaring issue is to anticipate changes and structure the service in such a way that changes will be discovered early to avoid change propagation. This is because, it is likely that services will be exposed to many unanticipated changes during its lifetime, and the support for unanticipated changes creates a gap for an important research goal. Two issues are involved here: (1) the assessment of the consequences of altering (or not) the functionality of the service, and (2) the identification of the service dependencies that are affected by the change. Traditionally, research on change impact analysis has mainly focused on technical aspects such as traceability analysis and software code change propagation. By contribution and extension of existing knowledge, the research presented in this thesis is exploratory in its nature, conducted through overarching emphasis of monitoring change propagation during service provisioning, with the understanding that software is now being consumed as services. The main goal is to advance the current state of change impact analysis by evolving a model-based framework for validating, analyzing and monitoring change propagation in our typical grid service provisioning environment (GUISET). The framework emanating from this endeavour consists of two associated formal models of change Impact Analysis Factor Adaptation mechanism, and a Fault and Failure Assumption Model for service provisioning in our GUISET grid environment. As a part of the empirical validation of the framework, we graphically represent the relationship between change and impact, indicating that, as the number of changes increased, the impact also increased. This is because the number of changes was affected by the number of dependent services. Therefore, if the dependencies were high, the number of changes will be high, and consequently the impact due to fault propagation will be high.
A Thesis Submitted in Fulfilment of the Requirements for the award of the Degree of Doctor of Philosophy (PhD) in Computer Science, Department of Computer Science Faculty of Science and Agriculture University of Zululand, 2009.
Grid-based Utility Infrastructure