Research On Large Legacy System Iterative Reengineering Based On Quantitative Methods

Legacy systems are systems which were developed many years ago. Normally, they are decayed but still important to users' business. After years of evolution, legacy systems have suffered from many problems, such as outdated technologies, decayed architectures, lack of documents and so on. These problems make the maintenance cost getting higher and higher. More and more users choose to reengineer legacy systems to new hardware/software platforms with modern design technologies. As one of the most important areas of software engineering, many applicative researches on software reengineering have been achieved in the last 30 years, such as design recovery, business rules extraction, legacy system migration/refactoring/reuse, test technologies, reengineering methodologies and so on. Iterative reengineering models become the dominant reengineering methodologies, because they can control project risk efficiently and make the legacy system keep serving in iterations.There are three major challenges in applying iterative reengineering processes to large complex legacy systems: The first is the development resource utilization. Traditional iterative reengineering models increase extra development cost and system complexisty which lower resource utilization ratio. The second is the performance of transition system in iterations. The large volume of data conversion will slow system performance. The last is the performance and network load in the distribute environment. The data transfer in the networking affects the performance of the distributed system a lot, if the service components are not deployed properly.This thesis is to solve these issues and achieved the following researches:Firstly, propose the Parallelism Iterative Reengineering Model (PIRM) which clarifies the transition architecture, iterative process and relative technologies. For large complex legacy system reengineering projects, it can achieve higher resource utilization ratio and higher transition system performance. The model defines goals and classifies applicable research achievements for every reengineering step which has an important guideline value. Secondly, propose Iteration Task Schedule Framework (ITSF) based on Concept Lattice to solve the task schedule issue in PIRM. According to transition architecture, the framework introduces a quantitative schedule method based on Access Strength (AS), and achieves the goals of performance optimization and resource utilization ratio increase.Thirdly, enhance the PIRM and proposes Distribute Performance Deployment Framework (DPDF) for legacy system Service Oriented Architecture (SOA) integration. DPDF introduces components Connectivity Strength with Parameter and Frequency (CSWPF) to measure the factors which affect the performance and proposes a hill-climbing Clustering Algorithm to solve the deployment issue for legacy system SOA integration.Chapter 1 introduces the background and research work in this thesis. Chapter 2 summarizes the research status and achievement in software reengineering fields. Chapter 3 presents the process and transition architecture of PIRM. Chapter 4 clarifies concept lattice based ITSF. Chapter 5 discusses the application issues of ITSF. Chapter 6 is the research extension of PIRM in SOA environment, which clarifies DPDF and relevant clustering algorithm. Chapter 7 describes the research application results of this thesis in two reengineering projects. Finally, chapter 8 summarizes the whole thesis and proposes the future work.