A Branch-and-Bound Algorithm for Autonomic Adaptation of Multi-cloud Applications

Adaptation is an important concern in cloud-based applications composed of services provided by different cloud providers since cloud services can suffer from Quality of Services(QoS) fluctuations. Other conditions that can also trigger an adaptation process at runtime are the unavailability of services or the violation of user-defined policies. Moreover, the detection and reaction on such changes must be done in an autonomic way, without the need of user intervention. This paper presents a dynamic adaptation approach for multi-cloud applications supported by a Branch-and-Bound (B&B) algorithm in order to optimize the adaptation process itself when selecting the services to be deployed within the application. Computational experiments comparing the B&B algorithm with another algorithm that evaluates all possible configurations for adapting an application showed that the B&B algorithm is faster than the previous version. This new algorithm brings benefits to the scalability of the adaptation process, which can deal with large configurations of multi-cloud applications composed by a plethora of cloud services.