Evaluating the Trade-off between DVFS Energy-savings and Virtual Networks Performance

Resumo—Data centers usually employ virtualization techniques coupled with other techniques, such as Dynamic Voltage and Frequency Scaling (DVFS), in order to reduce overall energy consumption. However, changes in processor frequency may impact the network performance, specially in metrics such as throughput and jitter. This paper evaluates the trade-off between changes in processor frequency and network performance. Our results show that there is an opportunity to save energy by up to 15%, through the processor frequency reduction. However, this reduction in frequency may increase the response time of applications by up to 70%, directly influencing the quality of experience (QoE).

[1]  Luiz Fernando Bittencourt,et al.  Power-aware virtual machine scheduling on clouds using active cooling control and DVFS , 2011, MGC '11.

[2]  Christine Mayap Kamga.  CPU Frequency Emulation Based on DVFS , 2012, 2012 IEEE Fifth International Conference on Utility and Cloud Computing.

[3]  Gregory Mone.  Redesigning the data center , 2012, CACM.

[4]  Giang Son Tran,et al.  Power-aware scheduler for virtualized systems , 2011, GCM '11.

[5]  Karthick Rajamani,et al.  Thermal response to DVFS: analysis with an Intel Pentium M , 2007, Proceedings of the 2007 international symposium on Low power electronics and design (ISLPED '07).

[6]  Christoph Meinel,et al.  Energy efficient scheduling of HPC-jobs on virtualize clusters using host and VM dynamic configuration , 2012, OPSR.

[7]  Stefanos Kaxiras,et al.  Poster: DVFS management in real-processors , 2011, ICS '11.

[8]  Qian Zhu,et al.  Power-Aware Consolidation of Scientific Workflows in Virtualized Environments , 2010, 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis.

[9]  Rajkumar Buyya,et al.  Energy Efficient Resource Management in Virtualized Cloud Data Centers , 2010, 2010 10th IEEE/ACM International Conference on Cluster, Cloud and Grid Computing.

[10]  Karsten Schwan,et al.  VirtualPower: coordinated power management in virtualized enterprise systems , 2007, SOSP.

[11]  Courtney Humphries,et al.  Towards power efficient consolidation and distribution of virtual machines , 2010, ACM SE '10.

[12]  Vikram A. Saletore,et al.  Evaluating network processing efficiency with processor partitioning and asynchronous I/O , 2006, EuroSys.

[13]  Rajkumar Buyya,et al.  CloudSim: a toolkit for modeling and simulation of cloud computing environments and evaluation of resource provisioning algorithms , 2011, Softw. Pract. Exp..

[14]  Gabriel Mateescu.  Overcoming the processor communication overhead in MPI applications , 2007, SpringSim '07.