CHASE: An Autonomic Service Engine for Cloud Environments

The rapidly spreading cloud computing paradigm delegates to the network the provision of most resources, even those strictly linked to hardware as storage and CPU time. This approach enables the development of applications which may exploit a variable amount of resources in a flexible way, so as to satisfy the actual load of requests coming from the users. A side effect of such flexibility is that optimization has to be more focused on user-perceived performance indexes than on resource usage. This paper takes a step in this direction, presenting the design and development of CHASE, an autonomic engine designed to optimize the scheduling of virtual machines in a cloud environment. The paper illustrates the CHASE architecture and its application in two different contexts: in PerfCloud, an environment for IaaS provision based on cloud and grid integration, and inside Cloud@Home, a project whose objective is to build a cloud using volunteer-based resources. Some preliminary experimental results based on HPC applications are presented.