Authors: Pan Yi Byrav Ramamurthy
Publish Date: 2015/12/10
Volume: 31, Issue: 3, Pages: 418-431
Abstract
Cloud computing enables the provisioning of resources in a reliable and ondemand manner With the increasing importance of the network bandwidth in the Cloud environment the networking related resources need to be optimally allocated together with the traditional Cloud computing resources In addition the significant growth of the global data center traffic raises the challenge of supporting demands with large bandwidth requirements for the Cloud provider In our paper we consider the networkefficient virtualized cloud infrastructure provisioning problem in IP over elastic optical network IPoverEON based on the data center as a service model The elastic optical network is adopted to provide spectrum and costefficient networking resources for large bandwidth requests in our work We develop mixed integer linear programming formulations to construct the mathematic model for this problem and propose a costoptimized heuristic to solve this problem To investigate the cost and blocking rate for the served demands different modulation formats are compared in the EON layer and the sliceable bandwidthvariable transponders and optical traffic grooming technology are considered The experimental results show that different modulation formats that are adopted in the EON layer will have different impacts on the total cost and demand blocking rate for the same data set size Also the use of SBVT will reduce the total cost no matter which modulation format is adopted and the reduction is related to the bandwidth requirement of the demandsThis work is supported by the National Science Foundation under Grant Nos CNS1040765 and CNS1345277 and Department of Energy DOE Project Award No DESC0001277 We thank the Holland Computing Center in UNL for providing the computing resources to conduct our experiments
Keywords: