Modelling and Analysis of Migration Policies for Autonomic Management of Energy Consumption in Cloud via Petri-Nets

There are often low-level migration policies that are enforced by the platform providers to reduce the cost associated with the consumption of energy. There are also high-level policies produced by the enterprises that own the virtual machines. For example, enterprises can impose location-oriented policies to stop the execution of tasks in a given country. The policies can also be aimed at boosting the so-called green credentials of the company. An example of such policies is a time-based policy for migrating virtual machines to various locations during off-peak times. These policies are often implemented as a part of autonomic systems that run the Cloud. Some of these policies are formed from Boolean constraints while others involve comparison of the values of variables with given threshold values. As a result, policies that govern migration of virtual machines in Cloud can be very complex. In particular, there is a clear need to identify the most appropriate policy among a set of potential policies. In this paper, we shall present a method using Coloured Petri-nets for modelling of a wide range of such policies so as to simulate them. We propose a method for calculating the Energy Cost and Transmission Cost along an execution path produced by the Petrinets models. For every group of policies the cost can be calculated and hence suitable policies can be identified. The approach has been evaluated with the help of a case study.