A Multiparameter Analytical Model of the Physical Infrastructure of a Cloud-Based System

Abstract

The large popularity of services offered by cloud computing (CC) requires constant expansion of its physical infrastructure. At the same time, CC operators apply various mechanisms that enable the existing physical resources to be optimally used. Useful tools in the design and optimization process are analytical and simulation models. They allow information about the operation of CC to be obtained without the need to make changes to the physical infrastructure. This article presents and discusses a multiservice, multiparameter model of a CC physical infrastructure that provides services of the infrastructure-as-a-service (IaaS) type. In the proposed model, the following four elements are used to specify network settings and describe the demands necessary to activate a virtual machine: the number of processors, the capacity of RAM, the HDD capacity, and the required network bitrate. Such an approach gives the opportunity to describe accurately the use of resources in real servers. To verify and validate the proposed model, we developed and implemented a simulator of the physical infrastructure of CC. The results of the simulations confirm the validity of all the theoretical assumptions adopted in the model. The proposed model can also serve as a tool, in the form of an appropriate application, for determining the resources that are necessary to service calls at the required loss level.