Research On Component Reuse Of Legacy System In Cloud Migration

In the process of software life cycle, due to the change of runtime envi-ronment and application requirement itself, organizers need adjust legacy sys-tems to protect and improve existing investment constantly. As a new business computing model, the dominant position of cloud computing is now strengthen-ing gradually in software market. In order to take advantage of the superiority brought by cloud computing, legacy software systems need make changes ac-cording to this new computing model, so migrating legacy systems to cloud platform is necessary.When migrating legacy systems to cloud platform, different migration strategies have different characteristics and advantages, mapping different reuse schemes. This dissertation focuses on component reuse in the process of migrat-ing partial function logic of object-oriented legacy system to SaaS, a group of approaches are proposed. First of all, the implementation information of legacy systems can be obtained by parsing the source code or byte code of legacy sys-tems. Then taking the implementation information as input, modules of legacy system can be recovered by using recovery approach of function structure, then reusable components can be extracted with the help of recovery approach of layer structure, and the description files of reusable components can be created through analyzing the related source code. Finally, the match between reusable components and target services can be carried out by combing text similarity measure and structure similarity assessment, thus the candidate components in legacy system can be reused efficiently. Based on the proposed approaches above, an assistant migration tool is developed, and one experiment is conducted to verify the effectiveness of these approaches. The experiment results show that the tool can effectively extract the reusable components in legacy system and effectively realize match between reusable components and target services. Re-lated work has a positive role in migrating partial business logic of legacy sys-tem to SaaS.In this dissertation, the main contributions and innovations are as follows:1. Put forward a new algorithm for recovering function structureModule dependency graph with relationship type information (R-MDG) is proposed to describe the implementation of legacy system. Taking R-MDG as input data, an improved vertical clustering algorithm is designed to recover the module structure of legacy system. Comparing to the existing Bunch algorithm, specific improvements include initial partition of clustering, multi-objective evaluation, metric of coupling and cohesion. The function structure recovered by the improved algorithm is more accurate, meanwhile the efficiency of clustering is enhanced by avoiding the randomness of initial partition.2. Propose a new approach for reusable component extractionAn improved horizontal clustering algorithm is adopted to realize, layer partition. By considering relationship type information between classes and package features in the process of clustering, layer partition results will be more reasonable. Function structure and layer structure obtained by clustering are overlapped orthogonally in orthogonal architecture recovery approach, then the independent modules in middle layer will be extracted to form reusable compo-nents. The dominant classes of reusable component can be found according to the relationship between the related classes, then the description information can be obtained by analyzing the dominant classes.3. Propose a new approach for matching target services and reusable com-ponentsBased on the description information of target services and reusable com-ponents, text similarity measure and structure similarity assessment are adopted to discover the reusable components related to specific target services. In view of the granularity difference between reusable components and target services, an iteration strategy is adopted to realize the match between target services and combined components, which can improve the possibility of finding more accu-rate reusable component related to target service.4. Develop a assistant tool for legacy system migrationOn the basis of DependencyFinder, the relationship type information be-tween classes is recovered in the process of reverse engineering. The assistant tool achieves four core functions, including function structure recovery, layer structure recovery, reusable components discovery, match between reusable components and target services. The tool can help software developer realize the reuse of business logic in legacy system efficiently.