Cloud restriction solver: A refactoring-based approach to migrate applications to the cloud

Abstract

ContextThe migration of legacy systems to the Platform as a Service (PaaS) model provides several benefits, but also brings new challenges, such as dealing with the restrictions imposed by the service provider. Furthermore, factors such as time, training and the extensive reengineering activities make the migration process time consuming and error prone. Although there exist several techniques for partial or total migration of legacy applications to the cloud, only a few specifically address the resolution of these constraints. ObjectiveThis paper proposes a novel semi-automatic approach, called Cloud Restriction Solver (CRS), for migrating applications to a PaaS environment that avoids the cloud restrictions through user-defined refactorings. MethodsThe approach is supported by two open and extensible tools. The first one, called CRSAnalyzer, identifies the pieces of code that violate the restrictions of the chosen PaaS platform, while the second one, CRSRefactor, changes those pieces by equivalent cloud-enabled services. ResultsThe applicability of the proposed approach is presented by showing its instantiation for Google App Engine as an Eclipse plugin and by migrating three Java applications to that PaaS successfully. In addition, an instantiation for IBM Bluemix has been created and used to compare the migration of the same application using the developed tools for both cloud providers. ConclusionThe proposed approach fosters software reuse, is cloud-independent, and facilitates the migration of applications to PaaS platforms.