Section: New Results

Cloud Computing and Deployment

Participants : Elena Giachino, Saverio Giallorenzo, Claudio Guidi, Cosimo Laneve, Michael Lienhardt, Jacopo Mauro, Gianluigi Zavattaro.

Cloud application deployment

Configuration and management of applications in the cloud is a complex task that requires advanced methodologies and tools. A foundational study of the problem has been carried out in [21] where we have identified the critical tasks to be solved, quantified their computational complexity, and proposed simplifications to the problem with the idea of limiting the computational complexity at the cost of having approximated (but acceptable, in most cases) solutions. Our attention has been dedicated to the implementation of a tool for the efficient solution of one of these tasks, namely, the automatic planning of the management actions needed to properly configure a cloud application [17] . This tool, called Metis, has been already exploited in the realization of an integrated platform for the automatic deployment of the cloud application called Blender [39] as well as in the context of the ABS modeling language [37] in order to be able to support the automatic reasoning about deployment costs already during the early phases of application design and development. We have also performed a foundational study of the problem of reconfiguring an application instead of deploying it from scratch. Our foundational study allowed us, on the one hand, to quantify the computational complexity of the problem (PSpace-Complete) and, on the other hand, to precisely identify the source of such complexity (the presence of legacy components that cannot be re-deployed from scratch).

Cloud resource management

One of the main challenges in the management of cloud applications is the quantification of the computing resources needed by the applications to be deployed. More precisely, it is important to quantify upper bounds to the number of needed computing resources in order to either previously acquire them or have a precise quantification of the costs for executing an application. In [40] a static analysis technique is proposed that computes upper bounds of virtual machine usages in a concurrent language with explicit acquire and release operations of virtual machines. See the section on deadlock analysis for more details.