Analysis and Optimization of the Trade-off between QoE and Energy-Efficiency in Data centers

点击翻译按钮获取中文摘要

Energy-efficient operation of data centers is gaining more and more attention. However, saving energy usually leads to a degradation of the available computing resources, which results in reduced Quality of Service (QoS). The diminished QoS can in turn lead to poor Quality of Experience (QoE) for the end-users. Although several approaches already address the optimization of data centers either to ensure energy-efficiency or to resolve certain QoS constraints, the joint optimization of both objectives is not researched. Furthermore, QoE is supposed to enable a holistic understanding of the qualitative performance of networked communication systems and thus to complement the traditional, more technology-centric QoS perspective.This project focuses on quantifying and adjusting the trade-off between QoE and energy-efficiency in data centers. We consider the use case of Virtual Desktop Infrastructures (VDIs) that are considered to have an increasing impact in the following years. A VDI enables users to use very lightweight systems, so-called thin clients, whereby the actual operating system runs in a data center. This approach may lead to cost and energy savings due to economies of scale by better utilization of existing resources in the data center. However, if too many services are aggregated onto very few servers, the service requirements cannot be fulfilled or result in significantly increased processing times. As a result, QoS and therefore QoE suffers. Additional challenges arise, as VDIs provide the user an entire operating system including various applications, e.g. Office products, with different requirements. The project aims at classifying the application types running inside VDIs according to their functionality and possible requirements on the client device, transmission network, and data center. Detailed QoE models for specific application-classes in a VDI environment will be derived based on subjective user studies e.g. via crowdsourcing. Such a QoE model quantifies the relation between the end-user QoE and different influence factors on the entire transmission chain including the client, network, and data center side. In parallel, corresponding models for the energy-consumption in data centers will be defined based on existing studies. Assuming that all information on the entire transmission chain is available, this allows for deriving the optimal trade-off between QoE and energy-efficiency by adjusting, e.g., scheduling of service requests, hot and cold stand-by of servers or virtual machine (VM) placement approaches.The methodology used in the proposed project combines literature studies, measurements based on dedicated testbeds, and subjective user studies with optimization methods using discrete-event simulations and analytical methods, like queuing theory. The work described in this proposal will be done in collaboration with the Hanoi University of Science and Technology, Vietnam