No abstract

International audienceCloud computing is nowadays one of the most promising IT technologies, since it provides seemingly unlimited resources on demand at low costs. Hence, different types of applications have been migrated to IaaS environments, e.g. multi-tier (distributed) applications. However, in order to benefit from such characteristics, cloud configurations (i.e. virtual resource configurations) should be designed accordingly to the necessities of the applications. Furthermore, such configurations have to provide the required resources not only at the application deployment-time, but also during the whole application execution time. Hence, adaptive paradigms are required when designing solutions to cloud applications with dynamic resource requirements. Software Product Lines (SPLs) provide great flexibility and a high level of abstraction to describe complete system configurations. Even though SPLs are not commonly used to describe changes after an initial product (configuration) has been created, their inherent characteristics can enable producing the required virtual resource configuration to adapt applications after their initial deployment, i.e., at runtime. In this paper, we present an approach to create and adapt cloud configurations at the IaaS level by using SPLs. We focus on the architectural design of our solution as well as on the possible implementation challenges we could face

International audienceOne key factor for Cloud computing success is the resource flexibility it provides. Because of this characteristic, academia and industry have focused their efforts on making efficient use of cloud computational resources without having to sacrifice performance. One way to achieve this purpose is through the automatic adaptation of the computational capabilities of VMs according to their resource utilization and performance. In this paper we present the design and preliminary results of our resource adaptation solution, which proactively adapts VMs (memory-based vertical scaling) to maintain an expected performance. Our solution targets multi-tier applications deployed on Cloud environments, and its core resides in RLS-based resource and performance predictors. Our results show that our solution, when compared with VMs with larger and permanently allocated computational resources, is able to maintain expected performance while reducing resource waste