Multi-attribute based double auction for virtual machine trading in federated cloud environment

Multi-attribute based double auction for virtual machine trading in federated cloud environment

연구 분야: Cryptography

학회: Cluster Computing

By participating in cloud federation, different cloud providers aim to increase their availability and optimize resource utilization to adapt to the time-varying service demands of cloud customers. As demand for computational resources is increasing rapidly, efficient resource allocation and provisioning decisions must be made within the federation to optimize both the cost and revenue of cloud service providers(CSPs). We propose a multi-attribute double auction mechanism for virtual machine trading (MABDA-VMT), enabling optimal virtual machine (VM) allocation for job execution in a dynamic federated cloud environment. We conducted a trace-driven simulation to examine the performance of our proposed model using the trace data from Google Cluster Usage Trace. The VM allocation is based on various cloud service attributes such as operational cost, communication cost, resource availability, requirements, and trust value. The proposed MABDA-VMT mechanism maximizes the profit and social welfare of the federation by considering these different attributes in the auction for VM trading. The proposed mechanism is also validated and satisfies the essential double auction properties like budget balance, individual rationality, truthfulness, and computational efficiency. Further, experimental results have been compared with the existing mechanisms, showing that the proposed mechanism produces the best results.