Yuan Tian
ABSTRACT
Server farms are playing an important role in the Internet infrastructure today. However, the increasing power consumption of server farms makes them expensive to operate. Thus, how to reduce the power consumed by server farms has become a important research topic. Power can be thought as a resource of system, just like traditional resources, and we can manage power via improved resource management strategies. In recent studies on power management, the system is attached with multiple states of different power consumption, and by switching among these states, power consumption can be made proportional to the work load. As different job scheduling policies will result in different performance and power consumption, an optimized policy with power as a factor can achieve a better tradeoff between performance and power consumption. In this paper, we summarize some familiar power management policies and propose a novel model using Stochastic Reward Nets(SRN). Based on this model, we analyze the performance and power consumption of different power management policies, and propose a novel cost-aware job scheduling algorithm.
- Epa report on server and data center energy efficiency. Technical report, U. S. Environmental Protection Agency, 2007.Google Scholar
- E. Altman, U. Ayesta, and B. Prabhu. Optimal load balancing in processor sharing systems. In In Proc. of GameComm. Citeseer, 2008.Google Scholar
- L. Andrew, M. Lin, and A. Wierman. Optimality, fairness, and robustness in speed scaling designs. ACM SIGMETRICS Performance Evaluation Review, 38(1):37--48, 2010. Google ScholarDigital Library
- N. Bansal, H. Chan, and K. Pruhs. Speed scaling with an arbitrary power function. In Proceedings of the twentieth Annual ACM-SIAM Symposium on Discrete Algorithms, pages 693--701. Society for Industrial and Applied Mathematics, 2009. Google ScholarDigital Library
- L. Barroso. The price of performance. Queue, 3(7):48--53, 2005. Google ScholarDigital Library
- L. Barroso and U. Holzle. The case for energy-proportional computing. Computer, 40(12):33--37, 2007. Google ScholarDigital Library
- L. Barroso and U. Hölzle. The datacenter as a computer: An introduction to the design of warehouse-scale machines. Synthesis Lectures on Computer Architecture, 4(1):1--108, 2009.Google ScholarCross Ref
- A. Beloglazov, R. Buyya, Y. Lee, A. Zomaya, et al. A taxonomy and survey of energy-efficient data centers and cloud computing systems. Advances in Computers, 82:47--111, 2011.Google ScholarCross Ref
- L. Benini, A. Bogliolo, G. Paleologo, and G. De Micheli. Policy optimization for dynamic power management. Computer-Aided Design of Integrated Circuits and Systems, IEEE Transactions on, 18(6):813--833, 1999. Google ScholarDigital Library
- J. Bradley. Optimal control of a dual service rate m/m/1 production-inventory model. European journal of operational research, 161(3):812--837, 2005.Google Scholar
- J. Chase, D. Anderson, P. Thakar, A. Vahdat, and R. Doyle. Managing energy and server resources in hosting centers. In Proceedings of the eighteenth ACM symposium on Operating systems principles, pages 103--116. ACM, 2001. Google ScholarDigital Library
- G. Chen, W. He, J. Liu, S. Nath, L. Rigas, L. Xiao, and F. Zhao. Energy-aware server provisioning and load dispatching for connection-intensive internet services. In Proceedings of the 5th USENIX Symposium on Networked Systems Design and Implementation, pages 337--350. USENIX Association, 2008. Google ScholarDigital Library
- L. Chen, N. Li, and S. Low. On the interaction between load balancing and speed scaling. In ITA Workshop, 2011.Google Scholar
- Y. Chen, A. Das, W. Qin, A. Sivasubramaniam, Q. Wang, and N. Gautam. Managing server energy and operational costs in hosting centers. In ACM SIGMETRICS Performance Evaluation Review, volume 33, pages 303--314. ACM, 2005. Google ScholarDigital Library
- L. Chuang. Computer Network and Performance Evaluation of Computer System. Tsinghua University Press, Apr 2001.Google Scholar
- A. Gandhi, V. Gupta, M. Harchol-Balter, and M. Kozuch. Optimality analysis of energy-performance trade-off for server farm management. Performance Evaluation, 2010. Google ScholarDigital Library
- A. Gandhi, M. Harchol-Balter, and I. Adan. Server farms with setup costs. Performance Evaluation, 2010. Google ScholarDigital Library
- R. German, C. Kelling, A. Zimmermann, and G. Hommel. Timenet: a toolkit for evaluating non-markovian stochastic petri nets. Performance Evaluation, 24(1):69--87, 1995. Google ScholarDigital Library
- B. Guenter, N. Jain, and C. Williams. Managing cost, performance, and reliability tradeoffs for energy-aware server provisioning. In INFOCOM, 2011 Proceedings IEEE, pages 1332--1340. IEEE, 2011.Google ScholarCross Ref
- M. Gupta, S. Grover, and S. Singh. A feasibility study for power management in lan switches. In Network Protocols, 2004. ICNP 2004. Proceedings of the 12th IEEE International Conference on, pages 361--371. IEEE, 2004. Google ScholarDigital Library
- S. Kaxiras and M. Martonosi. Computer architecture techniques for power-efficiency. Synthesis Lectures on Computer Architecture, 3(1):1--207, 2008. Google ScholarCross Ref
- M. Lin, A. Wierman, L. Andrew, and E. Thereska. Dynamic right-sizing for power-proportional data centers. In INFOCOM, 2011 Proceedings IEEE, pages 1098--1106. IEEE, 2011.Google ScholarCross Ref
- D. Meisner, B. Gold, and T. Wenisch. Powernap: eliminating server idle power. ACM SIGPLAN Notices, 44(3):205--216, 2009. Google ScholarDigital Library
- D. Palomar and M. Chiang. A tutorial on decomposition methods for network utility maximization. Selected Areas in Communications, IEEE Journal on, 24(8):1439--1451, 2006. Google ScholarDigital Library
- E. Pinheiro, R. Bianchini, E. Carrera, and T. Heath. Load balancing and unbalancing for power and performance in cluster-based systems. In Workshop on Compilers and Operating Systems for Low Power, volume 180, pages 182--195, 2001.Google Scholar
- A. Wierman, L. Andrew, and A. Tang. Power-aware speed scaling in processor sharing systems. In INFOCOM 2009, IEEE, pages 2007--2015. IEEE, 2009.Google ScholarCross Ref
- F. Yao, A. Demers, and S. Shenker. A scheduling model for reduced cpu energy. In Foundations of Computer Science, 1995. Proceedings., 36th Annual Symposium on, pages 374--382. IEEE, 1995. Google ScholarDigital Library
Index Terms
- Modeling and analyzing power management policies in server farms using stochastic Petri nets
Recommendations
Demand-aware power management for power-constrained HPC systems
CCGRID '16: Proceedings of the 16th IEEE/ACM International Symposium on Cluster, Cloud, and Grid ComputingAs limited power budget is becoming one of the most crucial challenges in developing supercomputer systems, hardware overprovisioning which installs larger number of nodes beyond the limitations of the power constraint determined by Thermal Design Power ...
Reliability-Aware Speed Control Policy for Energy Reduction in Server Farms
GREENCOM '12: Proceedings of the 2012 IEEE International Conference on Green Computing and CommunicationsServer farms are playing an important role in large-scale computing infrastructure today. However, the increasing energy consumption of server farms makes them expensive to operate. There are many technologies aiming at reducing the power consumption of ...
Modelling performance and power consumption of utilisation-based DVFS using M/M/1 queues
e-Energy '16: Proceedings of the Seventh International Conference on Future Energy SystemsDynamic Voltage and Frequency Scaling (DVFS) has become a key technique to reduce power consumption at times of low processor (CPU) utilisation. Reducing CPU frequency usually results in degraded services' performance leading to Service Level Agreement (...
Reviews
Hui Liu
A server farm or data center is a large collection of computer servers that provides services for enterprise and personal users. Many server farms run 24/7 and consume large amounts of electricity. It has been reported that federal servers and data centers in the US consumed 6.1 billion kilowatt hour (kWh) of electricity in 2006, at a total cost of $450 million [1]. Therefore, finding ways to reduce power consumption in server farms has become an important research topic. The authors of this paper investigated power management in server farms using stochastic Petri nets. The authors first reviewed data farm architecture, speed scaling policy, power-down policy, and job scheduling policy. For traditional load balancing strategies, energy-saving job scheduling assigns jobs to fewer servers. In this case, servers have more chances to turn off, which increases response time. By considering power consumption as a metric, a new model was proposed that considered more system states and policies. A cost model and an optimal global scheduling problem were also studied, and new theoretical results were obtained. In the end, numerical experiments focused on the average power consumption and response time. From these experiments, we can see when we should use load balancing policy and when we should use load concentration policy. The results are interesting and practical to data centers. The main contribution of this paper is that the authors proposed a more responsive model using stochastic Petri nets. The model considered more system states, such as power on, power off, sleep, and speed scaling, and several management policies. The second contribution is that the authors designed a cost-aware job scheduling algorithm. The authors also reviewed data center architecture, speed scaling, job scheduling, and different management policies. The information and method used in this paper will be helpful to other researchers. Online Computing Reviews Service
Access critical reviews of Computing literature here
Become a reviewer for Computing Reviews.
Comments