Optimistic virtual machine placement in cloud data centers using queuing approach

Cloud computing gives many beneficial services to share large scale of information, storage resources, computing resources, and provide knowledge for research. Cloud users deploy their own applications and related data on a pay-as-you-go basis. Virtual machines (VMs) usually host these data-intensive applications. The performance of these applications often depends on workload types I/O data-intensive or I/O computation, workload volume, CPU attributes on computing nodes, Virtual machines and the network. Therefore, the application jobs in the workload have different completion times based on the VM placement decision and large data retrieval. The main contribution of this thesis to gain high performance for the applications executed on the cloud by minimizing the completion time, minimizing the production cost and maximizing the throughput of cloud links. To provide a solution for minimizing the overall jobs' completion time (computing time as well as data transferring time) in both static and dynamic workloads, we propose VMs placement algorithm that considers computation resources, Quality of Service (QoS) metrics and virtual machine status and I/O data with priority based probability queuing model. The results obtained by the proposed methodology shows that the proposed optimal VM placement algorithm has a reduced processing cost and completion time compared with the traditional algorithms such as FCFS and priority scheduling. (C) 2018 Elsevier B.V. All rights reserved.