A Decade Time-Lapse of Cloud Performance and Dependability Modeling: Performability Evaluation Framework

A plethora of works has been published over the past decade, contributing in invaluable ways to the area of cloud performance and dependability modeling. As Cloud Computing became a new emerging technology aimed at large-scale resource sharing and service-oriented computing, their attention was spread over capacity planning, joint optimization of cost and capacity, analysis of failure characteristics of servers, failure and recovery behaviors, cloud service reliability, service availability and provisioning response delays, availability of a large-scale clouds with multiple classes of server pools, migration of physical machines among pools, cloud service performance, QoS measurements, simultaneous analysis of performance, availability, and power consumption, as well as end-to-end performability analysis of a cloud service. In this paper, we present a decade time-lapse of such research efforts, with a focus on the use of a class of Stochastic Petri Nets with reward structures as an integrated part of modeling, known as Stochastic Reward Nets. In addition, we propose a modeling framework for performability evaluation of a cloud service, which reveals considerable potential for further research in this area.