Administration autonome pour systèmes complexes de grande taille – SelfXL

Distributed systems are becoming more and more complex due to their heterogeneous architectures (on the physical and software level) in which resources are numerous and whose availability evolves at runtime. This complexity and dynamicity requires permanent adaptation, from the application to the system level, and requires to automate the adaptation process. The essence of autonomic computing is system self-management, freeing administrators and developers of low-level task configuration and management whilst delivering an optimized solution. In an autonomic system, the human operator defines general high level policies and rules that serve as an input for the adaptation process. These policies can be categorized into four functional classes : self-configuration, self-healing, self-optimization, self-protection. Since 2001 (IBM proposal), technical problems of self-management have been already studied (e.g., decision making) and sometimes solved (e.g., late binding, architectural reconfiguration or service monitoring). However, a major remaining challenge is to develop self-management of large-scale and complex systems in a efficient way. How can we address the autonomic management policies? How can we put the technical self-management solution fragments together in a systematic way? The partners of the SelfXL project seek to acquire new processes in the field of the autonomic computing, to investigate further the required components which would allow the implementation (language, runtime) of complex and large-scale autonomic system. The scope of this project encompasses any system that has a high software complexity (distributed, size of code etc.) and is large-scale in terms of size and heterogeneity of resources and software. Systems to be targeted range from cluster computing to embedded systems, including legacy software. Among the ANR project classifications, this project is an industrial project.