Mechanism and architecture for the migration of service implementation during traffic peaks

Abstract

Service-Oriented Architecture has been widely applied in enterprise computing systems for software-enabled services. However, cost efficiency and scalability requirements have moved the execution environment towards the cloud domain. Hybrid approaches have emerged, which utilise both enterprise and cloud domains in order to balance between the cost of service execution and the provided Quality of Service (QoS) for end users. This paper presents a migration, monitoring and load-balancing mechanism and architecture for scaling services between the enterprise and cloud domains during traffic peaks. The argued benefit of the proposal is the automation of the service-migration process and improvement of the QoS. A prototype system is presented as a proof of the conceptual architecture. The performance results in a hybrid cloud environment indicate that service implementation can be migrated and load can be balanced within 200 ms. Furthermore, the mechanism can improve the QoS for end users during traffic peaks. Our approach differs from existing proposals by focusing on the migration of service implementation, instead of the migration of service as part of a virtual machine.