Energy Management of Server Systems for Data Centers Using Dynamic Voltage-Frequency Scaling
0
조회
0
다운로드
7
인용
0
참고문헌
키워드
초록
Power consumption in data centers with large number of servers is being issued as energy costs increase. To reduce the energy costs, technologies such as DVFS are adopted to server systems. In this paper, we measure a multi-core server system’s power consumption running an HTTP server. To identify the characteristics of multi-core servers’ power consumption, we examined CPU-intensive and I/O-intensive synthetic workloads. Experimental results show that we can use DVFS for energy efficient server operation for CPU-intensive loads, while parallelizing the interrupt processing helps saving energy for I/O-intensive workloads.
Abstract
Power consumption in data centers with large number of servers is being issued as energy costs increase. To reduce the energy costs, technologies such as DVFS are adopted to server systems. In this paper, we measure a multi-core server system’s power consumption running an HTTP server. To identify the characteristics of multi-core servers’ power consumption, we examined CPU-intensive and I/O-intensive synthetic workloads. Experimental results show that we can use DVFS for energy efficient server operation for CPU-intensive loads, while parallelizing the interrupt processing helps saving energy for I/O-intensive workloads.
참고문헌 (14)
- M. Weiser, "Scheduling for Reduced CPU Energy", Proc. of the 1st USENIX Conf. on Operating Systems Design and Implementation, pp.13~23 (1994)
- C. Lefurgy, "Energy Management for Commercial Servers", IEEE Computer, 36(12), pp.39~48 (2003)
- R. Nathuji, "Feedback Driven QoS-Aware Power Budgeting for Virtualized Servers", Proc. of the Int'l Workshop On Feedback Control Implementation and Design in Computing Systems and Networks (2009)
- R. Bianchini, "Power and Energy Management for Server Systems", IEEE Computer, 37(11), pp.68~76 (2004)
- M.-a. Nasro, "Towards Dynamic Voltage Scaling in Real-Time Systems - A Survey", International Journal of Computer Sciences and Engineering Systems, 1(2), pp.93~103 (2007)
- J.-J. Chen, "Energy-Efficient Scheduling for Real-Time Systems on Dynamic Voltage Scaling (DVS) Platforms", Proc. of IEEE Int'l Conf. on Embedded and Real-Time Computing Systems and Applications, pp.28~38 (2007)
- T. Horvath, "Dynamic Voltage Scaling in Multitier Web Servers with End-to-End Delay Control", IEEE Transactions on Computers, 56(4), pp.444~458 (2007)
- T. Imada, "Power Management of Distributed WebServers by Controlling Server Power State andTraffic Prediction for QoS", Proceedings ofIEEE Int'l Symposium on Parallel andDistributed Processing, pp.1~8 (2008)
- Y. Wang, "Power-Efficient Response Time Guaranteesfor Virtualized Enterprise Servers", Proc. ofIEEE Real-Time Systems Symposium, pp.303~312 (2008)
- S. Chen, "Blackbox Prediction of the Impact of DVFS on End-to-End Performance of Multitier Systems", ACM SIGMETRICS Performance Evaluation Review, 37(4), pp.59~63 (2010)
- C. Santana, "Load Forecasting Applied to Soft Real-TimeWeb Clusters", Proceedings of the 2010 ACMSymposium on Applied Computing, pp.346~350 (2010)
- W.L. Bircher, "Predictive Power Management for Multi-Core Processors", Proc. of ISCA Workshop on Energy Efficient Design (2010)
- E. Seo, "Energy efficient scheduling of real-time tasks on multicore processors", IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, 19(11), pp.1540~1552, IEEE
- "4.2.1 Enhanced Intel® SpeedStep® Technology", In Intel® Core™ i7-800 and i5-700 Desktop Processor Series Datasheet, ~