Towards a Self-Adaptive Data Management System for Cloud Environments

International audienceCloud computing is an increasingly popular paradigm that gained interest from both scientific community and industry. As data volumes processed by applications running on clouds increase, the need for efficient and secure data management emerges as a crucial requirement. More specifically, storage systems intended for very large scales have to address a series of challenges, such as a scalable architecture, data location transparency or high throughput under concurrent accesses, requirements that come with a major drawback: the complexity of configuring and tuning the system's behavior. Such challenges can be overcome if the system is outfitted with a set of self-management components that enable an autonomic behavior. They heavily relies on introspection mechanisms, which play the crucial role of exposing the system's behavior accurately and in real time. This PhD research focuses on enhancing a distributed data-management system with self-management capabilities, so that it can meet the requirements of the Cloud storage services in terms of data availability, reliability and security. We focus on the case of BlobSeer, a system designed to store massive data, while leveraging a large-scale deployment and heavy data-access concurrency