Development of a simulation-based tool suite for modeling QoS/QoE versus cost trade-offs for cloud-deployed distributed software systems and services

点击翻译按钮获取中文摘要

Cloud computing's emergence offers opportunities by which modern public and private sector organizations can better manage the costs and efficiencies associated with acquiring and operating the computing resources required to support their operations. Many different variations of computing clouds though exist, spanning commodity commercial clouds, such as Amazon EC2 and Microsoft Azure, through to more tailored hybrid clouds, such as offered by RackSpace, to fully customized and tailored organizationally-held and operated fully private clouds. The quality of service (QoS) and customer-perceived quality of experience (QoE) levels can vary greatly depending on how a given organization's distributed software systems and/or services are deployed into which type of cloud instance. Costs also vary substantially as more tailored cloud deployments are required to meet the QoS/QoE that are desired. It is currently quite difficult to assessing the nature of these QoS/QoE versus cost trade-offs without engaging in costly and, typically, iterative trial-by-error deployments across the various options. This CRD project seeks to address this problem through developing a QoS/QoE versus cost cloud-deployment simulation tools suite that organizations can then use to cost-effectively explore the likely advantages and disadvantages of their various cloud deployment options, where this is inclusive of exploring "what-if" scenarios. For technical reasons, the simulator will be constructed as extensions to the widely used OMNet++ packet-level network simulator and designed to include: a) a cloud simulation module, b) an application-layer distributed system module, c) an incoming workload traffic module, and d) and an underlying inter- and intra-cloud bulk data transfer module. As such this CRD work seeks to extend the richness of QoS/QoE simulation detail currently provided in existing cloud simulation platforms in order to increase the accuracy of their QoS/QoE predictions. This will then allow real-world organizations to more informatively assess their cloud deployment options, particularly in terms of the resulting QoS/QoE versus cost trade-offs. The CRD is being done in collaboration with Referral SaaSquatch Inc. (Victoria, BC).