Font Size: a A A

An Application-oriented Processor Frequency Scaling Strategy For Energy Saving And Reliability Enhancement

Posted on:2017-11-13Degree:MasterType:Thesis
Country:ChinaCandidate:Z H WangFull Text:PDF
GTID:2428330590991625Subject:Software engineering
Abstract/Summary:PDF Full Text Request
As data centers flourish,research regarding data centers attracts people's attention.Among these kinds of research topics,how to reduce energy consumption in the data center has become the primary concern.For these research topics,management strategies for hardware resources have an impact on how the resources can be utilized in data centers.Especially,management strategies for processors are very important since processors consume about 42% energy of a data center.Therefore,how to reduce energy consumption of processors is the primary concern.Traditional dynamic voltage and frequency scaling technologies(DVFS)determine processor frequency based on processor utilization.They use high processor utilization as a predictor for user dissatisfaction,to which they react by increasing processor frequency.However,QoS requirements are not linearly related to processor utilization since they can vary from one service to another or one business scenario to another.When traditional CPU power management approaches can lose sight of user's actual QoS requirements,one opportunity for saving energy and enhancing reliability while meeting user's QoS requirements exists.Meanwhile,we should notice that user experience should not be sacrificed for energy reduction.User experience cannot be ignored in any condition.To save energy is to reduce energy consumption while guaranteeing user experience.Any energy efficient techniques which sacrifice user experience to save energy is not acceptable in practice.Therefore,in this paper,we address the importance of user experience.To help readers gain in-depth understanding of the importance of user experience,in this paper,we will present related work to show how models related to user experience can be built.In this paper,we propose application requirement driven(quality of service(QoS)driven)frequency scaling(ARD-FS)for energy-saving and reliability enhancement in static web servers,an adaptive application requirement(QoS)aware dynamic CPU frequency scaling technique.ARDFS is implemented on a Linux-based static web server.Compared with the default Linux CPU frequency controller,ARD-FS reduces the measured system-wide power consumption of the static web server by about 5% while meeting user's application requirements.Besides,ARD-FS is more reliable than conventional DVFS.For example,under some types of heavy workload,it is 17% more reliable than the default Linux CPU frequency controller in the static web server.
Keywords/Search Tags:Application Requirement Driven, QoS, Processor Frequency, Energy Reduction, Reliability
PDF Full Text Request
Related items