David Meisner
Luiz André Barroso
ABSTRACT
Much of the success of the Internet services model can be attributed to the popularity of a class of workloads that we call Online Data-Intensive (OLDI) services. These workloads perform significant computing over massive data sets per user request but, unlike their offline counterparts (such as MapReduce computations), they require responsiveness in the sub-second time scale at high request rates. Large search products, online advertising, and machine translation are examples of workloads in this class. Although the load in OLDI services can vary widely during the day, their energy consumption sees little variance due to the lack of energy proportionality of the underlying machinery. The scale and latency sensitivity of OLDI workloads also make them a challenging target for power management techniques.
We investigate what, if anything, can be done to make OLDI systems more energy-proportional. Specifically, we evaluate the applicability of active and idle low-power modes to reduce the power consumed by the primary server components (processor, memory, and disk), while maintaining tight response time constraints, particularly on 95th-percentile latency. Using Web search as a representative example of this workload class, we first characterize a production Web search workload at cluster-wide scale. We provide a fine-grain characterization and expose the opportunity for power savings using low-power modes of each primary server component. Second, we develop and validate a performance model to evaluate the impact of processor- and memory-based low-power modes on the search latency distribution and consider the benefit of current and foreseeable low-power modes. Our results highlight the challenges of power management for this class of workloads. In contrast to other server workloads, for which idle low-power modes have shown great promise, for OLDI workloads we find that energy-proportionality with acceptable query latency can only be achieved using coordinated, full-system active low-power modes.
Supplemental Material
- "AMD Family 10h Server and Workstation Processor Power and Thermal Data Sheet Rev 3.15" 2010.Google Scholar
- "Intel-Xeon Processor 5600 Series. Datasheet, Volume 1," 2010.Google Scholar
- L. A. Barroso and U. Hölzle, The Datacenter as a Computer. Morgan Claypool, 2009.Google Scholar
- L. A. Barroso, J. Dean, and U. Hölzle, "Web search for a planet: The google cluster architecture," IEEE Micro, vol. 23, no. 2, 2003. Google ScholarDigital Library
- L. A. Barroso and U. Hölzle, "The case for energy-proportional computing," Computer, vol. 40, no. 12, 2007. Google ScholarDigital Library
- D. Blaauw, S. Das, and Y. Lee, "Managing variations through adaptive design techniques," Tutorial at International Solid-State Circuits Conference, 2010.Google Scholar
- E. V. Carrera, E. Pinheiro, and R. Bianchini, "Conserving disk energy in network servers," in Proceedings of the 17th annual international conference on Supercomputing, 2003. Google ScholarDigital Library
- J. S. Chase, D. C. Anderson, P. N. Thakar, A. M. Vahdat, and R. P. Doyle, "Managing energy and server resources in hosting centers," in Symposium on Operating System Principles, 2001. Google ScholarDigital Library
- V. Delaluz, M. Kandemir, N. Vijaykrishnan, A. Sivasubramaniam, and M. J. Irwin, "Hardware and software techniques for controlling DRAM power modes," IEEE Trans. Comput., vol. 50, no. 11, 2001. Google ScholarDigital Library
- Q. Deng, D. Meisner, T. F. Wenisch, and R. Bianchini, "MemScale: Active Low-Power Modes for Main Memory," in Architectural Support for Programming Languages and Operating Systems, 2011. Google ScholarDigital Library
- B. Diniz, D. Guedes, W. Meira, Jr., and R. Bianchini, "Limiting the power consumption of main memory," in phInternational Symposium on Computer Architecture, 2007. Google ScholarDigital Library
- M. Elnozahy, M. Kistler, and R. Rajamony, "Energy conservation policies for web servers," in phProceedings of the 4th USENIX Symposium on Internet Technologies and Systems, 2003. Google ScholarDigital Library
- X. Fan, C. S. Ellis, and A. R. Lebeck, "The synergy between power-aware memory systems and processor voltage scaling," in Workshop on Power-Aware Computing Systems, 2003. Google ScholarDigital Library
- X. Fan, W.-D. Weber, and L. A. Barroso, "Power provisioning for a warehouse-sized computer," in International Symposium on Computer Architecture, 2007. Google ScholarDigital Library
- S. Gurumurthi, A. Sivasubramaniam, M. Kandemir, and H. Franke, "DRPM: dynamic speed control for power management in server class disks," in International Symposium on Computer Architecture, 2010. Google ScholarDigital Library
- T. Heath, B. Diniz, E. V. Carrera, W. Meira, Jr., and R. Bianchini, "Energy conservation in heterogeneous server clusters," in Principles and Practice of Parallel Programming, 2005. Google ScholarDigital Library
- J. Janzen, "Calculating memory system power for DDR SDRAM," Micron DesignLine, vol. 10, no. 2, 2001.Google Scholar
- S. Kaxiras and M. Martonosi, Computer Architecture Techniques for Power-Efficiency. Morgan Claypool, 2009. Google ScholarDigital Library
- K. Lim, P. Ranganathan, J. Chang, C. Patel, T. Mudge, and S. Reinhardt, "Understanding and designing new server architectures for emerging warehouse-computing environments," in International Symposium on Computer Architecture, 2008. Google ScholarDigital Library
- D. Meisner, B. T. Gold, and T. F. Wenisch, "PowerNap: Eliminating server idle power," in Architectural Support for Programming Languages and Operating Systems, 2009. Google ScholarDigital Library
- D. Meisner and T. F. Wenisch, "Stochastic Queuing Simulation for Data Center Workloads," in Exascale Evaluation and Research Techniques Workshop, 2010.Google Scholar
- E. Pinheiro and R. Bianchini, "Energy conservation techniques for disk array-based servers," in International Conference on Supercomputing, 2004. Google ScholarDigital Library
- K. Rajamani, C. Lefurgy, S. Ghiasi, J. Rubio, H. Hanson, and T. W. Keller, "Power management solutions for computer systems and datacenters," in International Symposium on Low-Power Electronics and Design, 2008. Google ScholarDigital Library
- E. Schurman and J. Brutlag, "The user and business impact of server delays, additional bytes, and HTTP chunking in web search," Velocity, 2009.Google Scholar
- D. Snowdon, S. Ruocco, and G. Heiser, "Power Management and Dynamic Voltage Scaling: Myths and Facts," in Workshop on Power Aware Real-time Computing, 2005.Google Scholar
- S. Srinivasan, L. Zhao, B. Ganesh, B. Jacob, M. Espig, and R. Iyer, "Cmp memory modeling: How much does accuracy matter?" in Workshop on Modeling, Benchmarking and Simulation, 2009.Google Scholar
- N. Tolia, Z. Wang, M. Marwah, C. Bash, P. Ranganathan, and X. Zhu, "Delivering energy proportionality with non energy-proportional systems -- optimizing the ensemble," in HotPower, 2008. Google ScholarDigital Library
- VJ Reddi, Benjamin Lee, Trishul Chilimbi, and Kushagra Vaid, "Web Search Using Mobile Cores: Quantifying and Mitigating the Price of Efficiency," in International Symposium on Computer Architecture, 2010. Google ScholarDigital Library
- Willis Lang and Jignesh M. Patel and Srinath Shankar, "Wimpy Node Clusters: What About Non-Wimpy Workloads?" in Workshop on Data Management on New Hardware, 2010. Google ScholarDigital Library
- F. Xie, M. Martonosi, and S. Malik, "Intraprogram dynamic voltage scaling: Bounding opportunities with analytic modeling," ACM Trans. Archit. Code Optim, vol. 1, p. 2004, 2004. Google ScholarDigital Library
Index Terms
- Power management of online data-intensive services
Recommendations
Power management of online data-intensive services
ISCA '11Much of the success of the Internet services model can be attributed to the popularity of a class of workloads that we call Online Data-Intensive (OLDI) services. These workloads perform significant computing over massive data sets per user request but, ...
Power-efficient distributed scheduling of virtual machines using workload-aware consolidation techniques
There is growing demand on datacenters to serve more clients with reasonable response times, demanding more hardware resources, and higher energy consumption. Energy-aware datacenters have thus been amongst the forerunners to deploy virtualization ...
Who Is Your Neighbor: Net I/O Performance Interference in Virtualized Clouds
User-perceived performance continues to be the most important QoS indicator in cloud-based data centers today. Effective allocation of virtual machines (VMs) to handle both CPU intensive and I/O intensive workloads is a crucial performance management ...
Comments