Control-oriented multirate LPV modelling of virtualized service center environments

To reduce the operating and energy costs of IT systems, nowadays software components are executed in virtualized servers, where a varying fraction of the physical CPU capacity is shared among running applications. Further, in service oriented environments, providers need to comply with the Service Level Objectives (SLO) stipulated in contracts with their customers. The trade-off between energy consumption and quality of service guarantees can be formalized in terms of a constrained control problem. To effectively solve it, accurate models of the virtualized server dynamics are needed. In this paper, an LPV multirate modeling approach is proposed, and its suitability for the considered application is assessed on experimental data.