Williamson, CareyWang, MeaLi, Yang2014-02-032014-03-152014-02-032014http://hdl.handle.net/11023/1365The topic of network-aware job placement in datacenter environments has attracted lots of research attention recently. Since network bandwidth is sometimes a scarce resource in multi-tenant datacenters, several virtualization abstractions have been proposed to avoid oversubscription of datacenter networks, and provide predictable QoS performance for users. In my work, a simulation environment is built to study job placement and scheduling strategies in datacenter networks. In particular, I model the Temporally-Interleaved Virtual Cluster (TIVC) algorithm, and study its sensitivity to different job characteristics and how the configured network capacity affects the system performance. The simulation results indicate that careful management of a delay queue for pending jobs can help achieve high system utilization. Different workload models have a pronounced influence on the system performance. In addition, an anomaly is found under certain datacenter network configurations, wherein increasing the datacenter resources produces "worse" system performance. Further investigation provides an explanation for this counter-intuitive result.engUniversity of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.Education--SciencesNetwork GuaranteeDatacenterJob Placementmaster thesis10.11575/PRISM/27075