Abstract
Cloud computing is a paradigm that is transforming the computing industry and is receiving more attention from the research community. The incremental deployment of cloud services is particularly important in agile development of cloud services, where successive cloud service increments must be integrated into existing cloud service architectures. This requires dynamic reconfiguration of software architectures, especially in cloud environments where services cannot be stopped in order to apply reconfiguration changes. This paper presents a model-driven dynamic architecture reconfiguration process to support the integration of cloud services. Models are used to represent high-level architecture reconfiguration operations as well as adaptation patterns. Adaptation patterns allow us to describe reconfiguration operations independently of a specific cloud platform technology. On the other hand, model transformations are used: i) to support compatibility checking of increments; ii) to generate software adaptors that solve incompatibilities between architectures; and iii) to generate platform specific reconfiguration plans, that include reconfiguration actions to be applied on cloud service instances at runtime. The proposed process is illustrated with a dealer network system development example, where cloud services are deployed in an incremental way.
