Cost-based virtual machine scheduling resilient to network traffic anomalies for Data as a Service

Cloud computing services run on top of different abstraction levels, involving many actors playing alternate roles as customers or providers. Financial losses caused by faults are propagated throughout all the systems' levels. Mechanisms to reduce costs and losses are important in face of the large scale of cloud computing systems. In this paper, we propose a cost-based scheduling model integrated into the network traffic anomaly detection engine. The Data-as-a-Service (DaaS) business model was taken as a case study to evaluate the impact resulting from anomalous traffic. The results showed a significant reduction in costs and financial losses, with gains ranging from 15% to 26% of the conventional method's financial losses.