Cloud computing is characterized by a model in which computing resources are delivered as services in a pay-as-you-go manner, which eliminates the need for upfront investments, reducing the time to market and opportunity costs. Despite its benefits, cloud computing brought new concerns about provider dependence and data confidentiality, which further led to a growing trend on consuming resources from multiple clouds. However, building multi-cloud systems is still very challenging and time consuming due to the heterogeneity across cloud providers' offerings and the high-variability in the configuration of cloud providers. This variability is expressed by the large number of available services and the many different ways in which they can be combined and configured. In order to ensure correct setup of a multi-cloud environment, developers must be aware of service offerings and configuration options from multiple cloud providers. To tackle this problem, this thesis proposes a software product line-based approach for managing the variability in cloud environments in order to automate the setup and adaptation of multi-cloud environments. The contributions of this thesis enable to automatically generate a configuration or reconfiguration plan for a multi-cloud environment from a description of its requirements. The conducted experiments aim to assess the impact of the approach on the automated analysis of feature models and the feasibility of the approach to automate the setup and adaptation of multi-cloud environments.
- Directeurs de thèse : Laurence DUCHIEN, Walter RUDAMETKIN - Rapporteurs : Philippe COLLET, Camille SALINESI - Examinateurs : Helene COULLON, Patrick HEYMANS
Thesis of the team Spirals defended on 05/06/2018