Section: Application Domains

Cluster and Grid computing

A cluster is a group of coupled computers that work together closely through fast Local Area Networks. Clusters are usually deployed within one administration domain to improve performance (for scientific applications) or availability (e.g., for Internet services hosted by a data center) compared to a single computer configuration. A grid is a collaboration between different administrative domains that share a part of their infrastructure. It is composed of a set of nodes which could be of very different nature, like clusters or (low-bandwidth wide-area) networks of personal computers or super-computers. Architectures require permanent adaptation, from the application to the system level and calls for automation of the adaptation process. We focus on self-configuration and self-optimization functionalities across the whole software stack : from the lower levels (systems mechanisms such as distributed file systems for instance) to the higher ones (i.e. the applications themselves such as J2EE clustered servers or scientific grid applications).

This year we extended the Entropy framework to allow the implementation of advanced scheduling policies though efficient cluster-wide context switches of virtualized jobs (i.e job embedded in one or several virtual machines) across a cluster [14] .

We continued to deal with self-administration of large distributed architectures with the partners of the ANR SelfXL project, see Sec.  8.2 . Furthermore, we finalized the MCITN Scalus European network in which we are involved. The project Scalus ("SCALing by means of Ubiquitous Storage") addresses the foundation for ubiquitous storage systems, which can be scaled with respect to multiple characteristics (capacity, performance, distance, security, ...), see Sec.  8.3 .