Virtually Timed Ambients: A Calculus for Resource Management in Cloud Computing

Cloud computing is a paradigm of distributed computing in which users share resources by storing data and executing processes in common data centers. A key factor for the success of this paradigm is virtualization technology, which represents the resources of an execution environment as a software layer, a so-called virtual machine. Virtualization allows to share existing hardware and software resources, improves security by providing isolation of different users, which share the same resource, and enables dynamic assignment of resources according to the demand of the user. The sharing of resources creates business drivers which make cloud computing an economically attractive model for deploying software. This thesis introduces the calculus of virtually timed ambients, a formal model of hierarchical locations for execution with explicit resource provisioning. This calculus is based on the well-known calculus of mobile ambients and motivated by the use of nested virtualization in cloud computing applications. The investigation of cloud computing from the point of view of process calculi provides a formal specification of the subject, which is necessary in order to develop executable models for analysis and optimization. The main contributions of this thesis are the definition of the calculus of virtually timed ambients, and the reasoning about its essential characteristics. In order to enable static analysis we enhance the calculus with a type system. Furthermore, we define a modal logic and a corresponding model checker, which we deploy in the definition of resource-awareness of virtually timed ambients, enabling dynamic self management of processes. Lastly, we present virtually timed ambients as a framework to analyse virtualization in cloud computing utilizing a prototype implementation. All concepts are illustrated by examples.