A Qos-Driven Approach to the Cloud Service Addressing Attributes of Security

Recently, cloud computing has been widely used by relying on its powerful resource integration and computing abilities. In the cloud computing system (CCS), the quality of service (QoS) is an important service evaluation criterion from provider and client perspectives, which directly affects the client experience and profit of the cloud providers. Thus, a precise evaluation of the QoS can help the cloud provider develop reasonable resource allocation strategies for improving the client experience. The performance metric is usually adopted to quantify QoS. Many approaches and methods for evaluating performance have been widely studied. However, another important metric, i.e., security, does not receive adequate attention in the evaluation of QoS. More importantly, security also has serious effects on the performance metric, that is, complex security-performance (S-P) correlations. To address these issues, this paper first builds a Markov model to analyze and assess the security of the CCS that captures two critical security factors, i.e., malicious attacks and the security protection mechanism. Then, a hierarchical modeling approach is presented to flexibly build the connection between security and the service performance. Finally, we propose a correlation metric to quantify random service performance. This correlation metric comprehensively considers the effect of the security factors and thus becomes more realistic and precise. The experimental results reveal the dynamic change of performance caused by the security factors and demonstrate the important S-P correlation. Therefore, security cannot be ignored in the modeling and evaluation of the QoS metric.