Fast provisioning of virtual machines in a distributed cloud

One of the primary functions of the Infrastructure as a Service (IaaS) cloud is to allocate the virtual machine (VM) resources upon users´ request. Cloud computing centers face the challenge of provisioning diverse VM instances in a short time. In addition, when a cloud service provider (CSP) operates a geographically diversified cloud that consists of large number of small data centers, the VM provisioning problem becomes even more complex due to the limits of network bandwidth and latency among the data centers. In this paper, we propose a fast virtual machines provisioning system in a distributed cloud environment. We implement a hierarchical system structure on the basis of the different network conditions inside and outside of data centers. In this system, we comprehensively use several methods to speed up the VM provisioning, including multilayered VM image structure, peer-to-peer (P2P) techniques, caching image on local and a cost aware VM placement scheduler. We implement the system on a real distributed cloud environment, and the performance of the presented methods is evaluated.