Autonomic Computing


In our research group, autonomic computing has been successfully applied to structured parallel applications: behavioural skeletons, control theoretic adaptation strategies and their application in DaSP contexts are successful stories. Behavioural skeletons, developed within the FP6 EU projects CoreGRID and GridCOMP, associate autonomic managers taking care of non functional concerns in algorithmic skeletons modelling common parallel patterns and have been demonstrated successful in different kind of parallel applications. Different kind of autonomic management strategies have been investigated in the field of DaSP (Data Stream Processing) leading to interesting results supporting dynamic adaptation of parallel DaSP pattern execution. Eventually, autonomic managers have been added to FastFlow patterns to adapt pattern execution to performance and energy related requirements (deadlines, power budget) varying in time.

Comments are closed