# Mercury and Freon: Temperature Emulation and Management for Server Systems

Save this PDF as:

Size: px
Start display at page:

## Transcription

3 Disk Platters Motherboard Power Supply Inlet AC Disk Air Power Supply Air Disk Air Downstream Power Supply Air Downstream Disk Shell CPU Void Space Air Power Supply Air Void Space Air Machine 1 Machine 2 Machine 3 Machine 4 CPU Air CPU Air Downstream Disk Air CPU Air Exhaust Cluster Exhaust (a) (b) (c) Figure 1. Example intra-machine heat-flow (a), intra-machine air-flow (b), and inter-machine air-flow (c) graphs. where Q transfer,1 2 is the amount of heat transferred between objects 1 and 2 or between object 1 and its environment during a period of time, T 1 and T 2 are their current temperatures, and k is a constant that embodies the heat transfer coefficient and the surface area of the object. The k constant can be approximated by experiment, simulation, or rough numerical approximation using typical engineering equations. However, in reality, k can vary with temperature and air-flow rates. We assume that k is constant in our modeling because the extra accuracy of a variable k would not justify the complexity involved in instantiating it for all temperatures and rates. Energy equivalent. The heat produced by a component essentially corresponds to the energy it consumes, so we define it as: Q component = P (utilization) time (3) where P (utilization) represents the average power consumed by the component as a function of its utilization. For the components that we have studied in detail so far, a simple linear formulation has correctly approximated the real power consumption: P (utilization) = P base + utilization (P max P base ) (4) where P base is the power consumption when the component is idle and P max is the consumption when the component is fully utilized. Researchers have used similar high-level formulations to model modern processors and server-style DRAM subsystems [8]. However, our default formulation can be easily replaced by a more sophisticated one for components that do not exhibit a linear relationship between high-level utilization and power consumption. For example, we have an alternate formulation for Q CP U that is based on hardware performance counters. (Although our descriptions and results assume the default formulation, we do discuss this alternate formulation in Section 2.3.) Heat capacity. Finally, since pressures and volumes in our system are assumed constant, the temperature of an object is directly proportional to its internal energy. More formally, we define the object s temperature variation as: T = 1 Q (5) m c where m is the mass of the object and c is its specific heat capacity. 2.2 Inputs and Temperature Emulation Mercury takes three groups of inputs: heat- and air-flow graphs describing the layout of the hardware and the air circulation, constants describing the physical properties and the power consumption of the hardware components, and dynamic component utilizations. Next, we describe these groups and how Mercury uses them. Graphs. At its heart, Mercury is a coarse-grained finite element analyzer. The elements are specified as vertices on a graph, and the edges represent either air-flow or heat-flow properties. More specifically, Mercury uses three input graphs: an inter-component heat-flow graph, an intra-machine air-flow graph, and possibly an inter-machine air-flow graph for clustered systems. The heat-flow graphs are undirected, since the direction of heat flow is solely dependent upon the difference in temperature between each pair of components. The air-flow graphs are directed, since fans physically move air from one point to another in the system. Figure 1 presents one example of each type of graph. Figure 1(a) shows a heat-flow graph for the real server we use to validate Mercury in this paper. The figure includes vertices for the CPU, the disk, the power supply, and the motherboard, as well as the air around each of these components. As suggested by this figure, Mercury computes a single temperature for an entire hardware component, which again is sufficient for whole-system thermal management research. Figure 1(b) shows the intra-machine airflow for the same server. In this case, vertices represent air regions, such as inlet air and the air that flows over the CPU. Finally, Figure 1(c) shows an inter-machine air flow graph for a small cluster of four machines. The vertices here represent inlet and outlet air regions for each machine, the cold air coming from the air conditioner, and the hot air to be chilled. The graph represents the ideal situation in which there is no air recirculation across the machines. Recirculation and rack layout effects can also be represented using more complex graphs. Constants. The discussion above has not detailed how the input graphs edges are labeled. To label the edges of the heat-flow graph and instantiate the equations described in the previous subsection, Mercury takes as input the heat transfer coefficients, the components specific heat capacities, and the components surface areas and masses. The air-flow edges are labeled with the fraction of air that flows from one vertex to another. For example, the airflow edge from the Disk Air Downstream vertex to the CPU Air vertex in Figure 1(b) specifies how much of the air that flows over the disk flows over the CPU as well. (Because some modern fans change speeds dynamically, we need a way to adjust the airflow fractions accordingly. The thermal emergency tool described in Section 2.3 can be used for this purpose.) Besides these con-

5 #!/bin/bash sleep 1 fiddle machine1 temperature inlet sleep fiddle machine1 temperature inlet 21.6 Figure 4. An example fiddle script. temperature on-line. For example, the user can simulate the failure of an air conditioner or the accidental blocking of a machine s air inlet in a data center by explicitly setting high temperatures for some of the machine inlets. Figure 4 shows an example fiddle script that simulates a cooling failure by setting the temperature of machine1 s inlet air to o C 1 seconds into the experiment. The high inlet temperature remains for seconds, at which point we simulate a restoration of the cooling. Fiddle can also be used to change air-flow or power-consumption information dynamically, allowing us to emulate multi-speed fans and CPU-driven thermal management using voltage/frequency scaling or clock throttling. We discuss this issue in Section 7. Mercury for modern processors. As we mentioned above, computing the amount of heat produced by a CPU using high-level utilization information may not be adequate for modern processors. For this reason, we have recently developed and validated a version of Mercury for the Intel Pentium 4 processor. In this version, monitord monitors the hardware performance counters and translates each observed performance event into an estimated energy [2]. (The software infrastructure for handling the Intel performance counters has been provided by Frank Bellosa s group.) To avoid further modifications to Mercury, the daemon transforms the estimated energy during each interval into an average power, which is then linearly translated into a low-level utilization in the [% = P base, 1% = P max] range. This is the utilization reported to the solver. 3. Mercury Validation We validated Mercury using two types of single-server experiments. The first type evaluates Mercury as a replacement for real measurements, which are difficult to repeat and prevent the study of thermal emergencies. In these validations, we compare the Mercury temperatures to real measurements of benchmarks running through a series of different CPU and disk utilizations. The second type of validations evaluate Mercury as a replacement for time-consuming simulators, which are typically not capable of executing applications or systems software. In these validations, we compare the Mercury (steady-state) temperatures for a simple server case against those of a 2D simulation of the case at several different, but constant component power consumptions. 3.1 Real Machine Experiments To validate with a real machine, we used a server with a single Pentium III CPU, 512 MBytes of main memory, and a SCSI 15K-rpm disk. We configured Mercury with the parameters shown in Table 1 and the intra-machine graphs of Figure 1(a) and 1(b). The thermal constants in the table were defined as follows. As in [12], we modeled the disk drive as a disk core and a disk shell (base and cover) around it. Also as in [12], we assumed the specific heat capacity of aluminum for both disk drive components. We did the same for the CPU (plus heat sink), representing the specific heat capacity of the heat sink. We assumed the specific heat capacity of FR4 (Flame Resistant 4) for the motherboard. The k constant for the different heat transfers was defined using calibration, as described below. We estimated the air fractions based on the layout of the components in our server. All the component masses were Component Property Value Unit Disk Platters Mass.336 kg J Specific Heat Capacity 896 Kkg (Min, Max) Power (9, 14) Watts Disk Shell Mass.55 kg J Specific Heat Capacity 896 Kkg CPU Mass.151 kg J Specific Heat Capacity 896 Kkg (Min, Max) Power (7, 31) Watts Power Supply Mass kg J Specific Heat Capacity 896 Kkg (Min, Max) Power (, ) Watts Motherboard Mass.718 kg J Specific Heat Capacity 1245 Kkg (Min, Max) Power (4, 4) Watts Inlet Temperature 21.6 Celsius Fan Speed 38.6 ft 3 /min From/To To/From k (W atts/kelvin) Disk Platters Disk Shell 2. Disk Shell Disk Air 1.9 CPU CPU Air.75 Power Supply Power Supply Air 4 Motherboard Void Space Air 1 Motherboard CPU.1 From To Air Fraction Inlet Disk Air.4 Inlet PS Air.5 Inlet Void Space Air.1 Disk Air Disk Air Downstream 1. Disk Air Downstream Void Space Air 1. PS Air PS Air Downstream 1. PS Air Downstream Void Space Air.85 PS Air Downstream CPU Air.15 Void Space Air CPU Air.5 Void Space Air Exhaust.95 CPU Air CPU Air Downstream 1. CPU Air Downstream Exhaust 1. From To Air Fraction AC Machine 1.25 AC Machine 2.25 AC Machine 3.25 AC Machine 4.25 Machine 1 Cluster Exhaust 1. Machine 2 Cluster Exhaust 1. Machine 3 Cluster Exhaust 1. Machine 4 Cluster Exhaust 1. Table 1. Constants used in the validation and Freon studies. measured; the CPU was weighed with its heat sink. The disk power consumptions came from the disk s datasheet, whereas we measured the CPU power consumptions. We also measured the power consumption of the motherboard without any removable components (e.g., the CPU and memory chips) and the power supply at 44 Watts. From this number, we estimated the power supply consumption at Watts, representing a 13% efficiency loss with respect to its -Watt rating. We ran calibration experiments with two microbenchmarks. The first set exercises the CPU, putting it through various levels of utilization interspersed with idle periods. The second does the same for the disk. We placed a temperature sensor on top of the CPU heat sink (to measure the temperature of the air heated up by the CPU), and measured the disk temperature using its internal sensor.

6 1 CPU Util Emulated Real 1 Disk Util Emulated Real Percent Utilization Temperature (C) Percent Utilization Temperature (C) Time (Seconds) Figure 5. Calibrating Mercury for CPU usage and temperature Time (Seconds) Figure 6. Calibrating Mercury for disk usage and temperature. 1 CPU Util Emulated Real 1 Disk Util Emulated Real Percent Utilization Temperature (C) Percent Utilization Temperature (C) Time (Seconds) Figure 7. Real-system CPU air validation Time (Seconds) Figure 8. Real-system disk validation. Figures 5 and 6 illustrate the component utilizations and measured temperatures during the experiments with the CPU and the disk, respectively. The figures also show the Mercury temperatures for the CPU air and the disk, after a phase of input calibrations based on the real measurements. The calibrations were easy to do, taking one of us less than an hour to perform. From these figures, we observe that Mercury is able to compute temperatures that are very close to the measurements for the microbenchmarks. The calibration allows us to correct inaccuracies in all aspects of the thermal modeling. After the calibration phase, Mercury can be used without ever collecting more real temperature measurements, i.e. Mercury will consistently behave as the real system did during the calibration run. To validate that Mercury is indeed accurate after calibration, we ran a more challenging benchmark without adjusting any input parameters. Figures 7 (CPU air) and 8 (disk) compare Mercury and real temperatures for the benchmark. The benchmark exercises the CPU and disk at the same time, generating widely different utilizations over time. This behavior is difficult to track, since utilizations change constantly and quickly. Despite the challenging benchmark, the validation shows that Mercury is able to emulate temperatures within 1 o C at all times (see the Y-axis on the right of Figures 7 and 8). For comparison, this accuracy is actually better than that of our real digital thermometers (1.5 o C) and indisk sensors (3 o C). We have also run validation experiments with variations of this benchmark as well as real applications, such as the Web server and workload used in Section 5. In these (less-challenging) experiments, accuracy was at least as good as in Figures 7 and Simulated Machine Experiments The other main use of the Mercury system is to replace complex simulators. Thus, in this set of validations, we use Fluent, a widely used commercial simulator [9]. The Fluent interface does not allow us to easily vary the power consumption of the simulated CPU, so we compare Mercury and Fluent in steady-state for a variety of fixed power consumptions. We modeled a 2D description of a server case, with a CPU, a disk, and a power supply. Once the geometry was entered and meshed, Fluent was able to calculate the heat-transfer properties of the material-to-air boundaries. Our calibration of Mercury involved entering these values as input, with a rough approximation of the air flow that was also provided by Fluent. The steady-state temperatures were then found by running Mercury and Fluent for given values of power supply, CPU, and disk power consumptions. Note that here we validated CPU temperatures, rather than CPU air temperatures as in the previous subsection. Since the server case was modeled in 2D, no air flows over the simulated CPU. Our validation results for 14 experiments, each one for a different combination of CPU and disk power consumptions, show that Mercury approximates the Fluent temperatures quite well (within.25 o C for the disk and.32 o C for the CPU). Note that Mercury achieves this high accuracy despite the fact that Fluent models many hundreds of mesh points, doing a complex analysis of

11 ing the Mercury code) or externally (via fiddle). We also plan to study the emulation of chip multiprocessors, which will probably have to be done in two levels, for each core and the entire chip. Our real implementation of Freon demonstrated that relatively simple policies and systems can effectively manage thermal emergencies. However, the current version of Freon also needs a few extensions. In particular, Freon needs to be extended to deal with multi-tier services and to include other policies. We are currently addressing these extensions, as well as considering ways to combine Freon with CPU-driven thermal management. Acknowledgements We would like to thank Frank Bellosa, Enrique V. Carrera, Steve Dropsho, Cosmin Rusu, and the anonymous reviewers for comments that helped improve this paper significantly. We are also indebted to Frank Bellosa, Andreas Merkel, and Simon Kellner for sharing their performance-counter infrastructure with us. Finally, we thank Enrique V. Carrera and Diego Nogueira who contributed to the early stages of this work. References [1] D. Anderson, J. Dykes, and E. Riedel. More than an Interface SCSI vs. ATA. In Proceedings of FAST, March 3. [2] F. Bellosa, S. Kellner, M. Waitz, and A. Weissel. Event-Driven Energy Accounting of Dynamic Thermal Management. In Proceedings of COLP, September 3. [3] D. Brooks and M. Martonosi. Dynamic Thermal Management for High-Performance Microprocessors. In Proceedings of HPCA, January 1. [4] J. Chase, D. Anderson, P. Thackar, A. Vahdat, and R. Boyle. Managing Energy and Server Resources in Hosting Centers. In Proceedings of SOSP, October 1. [5] Y. Chen, A. Das, W. Qin, A. Sivasubramaniam, Q. Wang, and N. Gautam. Managing Server Energy and Operational Costs in Hosting Centers. In Proceedings of Sigmetrics, June 5. [6] E. N. Elnozahy, M. Kistler, and R. Rajamony. Energy-Efficient Server Clusters. In Proceedings of PACS, February 2. [7] Ericsson Microelectronics. Reliability Aspects on Power Supplies. Technical Report Design Note 2, April. [8] W. Felter, K. Rajamani, T. Keller, and C. Rusu. A Performance- Conserving Approach for Reducing Peak Power Consumption in Server Systems. In Proceedings of ICS, June 5. [9] Fluent. Fluent: The Right Answer in CFD. index.htm. [1] E. Gansner, E. Koutsofios, and S. North. Drawing Graphs with Dot. [11] S. Gurumurthi, A. Sivasubramaniam, M. Kandemir, and H. Franke. DRPM: Dynamic Speed Control for Power Management in Server Class Disks. In Proceedings of ISCA, June 3. [12] S. Gurumurthi, A. Sivasubramaniam, and V. Natarajan. Disk Drive Roadmap from the Thermal Perspective: A Case for Dynamic Thermal Management. In Proceedings of ISCA, June 5. [13] T. Heath, A. P. Centeno, P. George, L. Ramos, Y. Jaluria, and R. Bianchini. Mercury and Freon: Temperature Emulation and Management for Server Systems. Technical Report DCS-TR-596, Department of Computer Science,, January 6, Revised July 6. [14] T. Heath, B. Diniz, E. V. Carrera, W. Meira Jr., and R. Bianchini. Energy Conservation in Heterogeneous Server Clusters. In Proceedings of PPoPP, June 5. [15] T. Icoz and Y. Jaluria. Numerical Simulation of Boundary Conditions and the Onset of Instability in Natural Convection Due to Protruding Thermal Sources in an Open Rectangular Channel. Numerical Heat Transfer, 48: , 5. [16] Y. Kim, S. Gurumurthi, and A. Sivasubramaniam. Understanding the Performance-Temperature Interactions in Disk I/O of Server Workloads. In Proceedings of HPCA, February 6. [17] Y. Li, D. Brooks, Z. Hu, and K. Skadron. Performance, Energy, and Thermal Considerations for SMT and CMP Architectures. In Proceedings of HPCA, February 5. [18] A. Merkel and F. Bellosa. Balancing Power Consumption in Multiprocessor Systems. In Proceedings of Eurosys, April 6. [19] J. Moore, J. Chase, and P. Ranganathan. ConSil: Low-Cost Thermal Mapping of Data Centers. In Proceedings of SysML, June 6. [] J. Moore, J. Chase, and P. Ranganathan. Weatherman: Automated, Online, and Predictive Thermal Mapping and Management for Data Centers. In Proceedings of ICAC, June 6. [21] J. Moore, J. Chase, P. Ranganathan, and R. Sharma. Making Scheduling Cool: Temperature-Aware Resource Assignment in Data Centers. In Proceedings of USENIX, April 5. [22] E. Pinheiro, R. Bianchini, E. Carrera, and T. Heath. Dynamic Cluster Reconfiguration for Power and Performance. In L. Benini, M. Kandemir, and J. Ramanujam, editors, Compilers and Operating Systems for Low Power. Kluwer Academic Publishers, August 3. Earlier version published in Proceedings of COLP 1, September 1. [23] M. Powell, M. Gomaa, and T. N. Vijaykumar. Heat-and-Run: Leveraging SMT and CMP to Manage Power Density Through the Operating System. In Proceedings of ASPLOS, October 4. [24] F. K. Price. Heat Transfer and Fluid Flow Data Books. Genium Publishing Corp., 199. [25] K. Rajamani and C. Lefurgy. On Evaluating Request-Distribution Schemes for Saving Energy in Server Clusters. In Proceedings of ISPASS, March 3. [26] E. Rohou and M. D. Smith. Dynamically Managing Processor Temperature and Power. In Proceedings of FDO, November [27] C. Rusu, A. Ferreira, C. Scordino, A. Watson, R. Melhem, and D. Mosse. Energy-Efficient Real-Time Heterogeneous Server Clusters. In Proceedings of RTAS, April 6. [28] L. Shang, L.-S. Peh, A. Kumar, and N. Jha. Characterization and Management of On-Chip Networks. In Proceedings of Micro, December 4. [29] R. Sharma, C. Bash, C. Patel, R. Friedrich, and J. Chase. Balance of Power: Dynamic Thermal Management for Internet Data Centers. IEEE Internet Computing, 9(1), January 5. [] K. Skadron, M. Stan, W. Huang, S. Velusamy, K. Sankaranarayanan, and D. Tarjan. Temperature-Aware Microarchitecture. In Proceedings of ISCA, June 3. [31] J. Srinivasan, S. Adve, P. Bose, and J. Rivers. The Case for Lifetime Reliability-Aware Microprocessors. In Proceedings of ISCA, June 4. [32] A. Weissel and F. Bellosa. Dynamic Thermal Management for Distributed Systems. In Proceedings of TACS, June 4. [33] W. Zhang. Linux Virtual Server for Scalable Network Services. In Proceedings of the Linux Symposium, July.

### Mercury and Freon: Temperature Emulation and Management for Server Systems

Mercury and Freon: Temperature Emulation and Management for Server Systems Taliver Heath Dept. of Computer Science taliver@cs.rutgers.edu Ana Paula Centeno Dept. of Computer Science anapaula@cs.rutgers.edu

### C-Oracle: Predictive Thermal Management for Data Centers

C-Oracle: Predictive Thermal Management for Data Centers Luiz Ramos and Ricardo Bianchini Department of Computer Science, Rutgers University {luramos,ricardob}@cs.rutgers.edu Abstract Designing thermal

### Temperature measurement in the Intel Core TM Duo Processor

Temperature measurement in the Intel Core TM Duo Processor Efraim Rotem Mobile Platform Group, Intel corporation Jim Hermerding Mobile platform Group, Intel corporation Cohen Aviad - Microprocessor Technology

### Case study: End-to-end data centre infrastructure management

Case study: End-to-end data centre infrastructure management Situation: A leading public sector organisation suspected that their air conditioning units were not cooling the data centre efficiently. Consequently,

### Windows Server 2008 R2 Hyper-V Live Migration

Windows Server 2008 R2 Hyper-V Live Migration Table of Contents Overview of Windows Server 2008 R2 Hyper-V Features... 3 Dynamic VM storage... 3 Enhanced Processor Support... 3 Enhanced Networking Support...

### Dynamic Power Variations in Data Centers and Network Rooms

Dynamic Power Variations in Data Centers and Network Rooms White Paper 43 Revision 3 by James Spitaels > Executive summary The power requirement required by data centers and network rooms varies on a minute

### Dynamic Power Variations in Data Centers and Network Rooms

Dynamic Power Variations in Data Centers and Network Rooms By Jim Spitaels White Paper #43 Revision 2 Executive Summary The power requirement required by data centers and network rooms varies on a minute

### Windows Server Performance Monitoring

Spot server problems before they are noticed The system s really slow today! How often have you heard that? Finding the solution isn t so easy. The obvious questions to ask are why is it running slowly

### Green HPC - Dynamic Power Management in HPC

Gr eenhpc Dynami cpower Management i nhpc AT ECHNOL OGYWHI T EP APER Green HPC Dynamic Power Management in HPC 2 Green HPC - Dynamic Power Management in HPC Introduction... 3 Green Strategies... 4 Implementation...

### OPTIMIZING SERVER VIRTUALIZATION

OPTIMIZING SERVER VIRTUALIZATION HP MULTI-PORT SERVER ADAPTERS BASED ON INTEL ETHERNET TECHNOLOGY As enterprise-class server infrastructures adopt virtualization to improve total cost of ownership (TCO)

### Raid Ayoub Tajana Simunic Rosing Computer Science and Engineering Department UC San Diego

Raid Ayoub Tajana Simunic Rosing Computer Science and Engineering Department UC San Diego Thermal stress challenges Market demand for high integration Market demand for high performance High power density

### Windows Server 2008 R2 Hyper-V Live Migration

Windows Server 2008 R2 Hyper-V Live Migration White Paper Published: August 09 This is a preliminary document and may be changed substantially prior to final commercial release of the software described

### Data Center Specific Thermal and Energy Saving Techniques

Data Center Specific Thermal and Energy Saving Techniques Tausif Muzaffar and Xiao Qin Department of Computer Science and Software Engineering Auburn University 1 Big Data 2 Data Centers In 2013, there

### Cloud Server. Parallels. An Introduction to Operating System Virtualization and Parallels Cloud Server. White Paper. www.parallels.

Parallels Cloud Server White Paper An Introduction to Operating System Virtualization and Parallels Cloud Server www.parallels.com Table of Contents Introduction... 3 Hardware Virtualization... 3 Operating

### Leveraging Thermal Storage to Cut the Electricity Bill for Datacenter Cooling

Leveraging Thermal Storage to Cut the Electricity Bill for Datacenter Cooling Yefu Wang1, Xiaorui Wang1,2, and Yanwei Zhang1 ABSTRACT The Ohio State University 14 1 1 8 6 4 9 8 Time (1 minuts) 7 6 4 3

### Managing Data Center Power and Cooling

White PAPER Managing Data Center Power and Cooling Introduction: Crisis in Power and Cooling As server microprocessors become more powerful in accordance with Moore s Law, they also consume more power

### PARALLELS CLOUD SERVER

PARALLELS CLOUD SERVER An Introduction to Operating System Virtualization and Parallels Cloud Server 1 Table of Contents Introduction... 3 Hardware Virtualization... 3 Operating System Virtualization...

### Cooling Capacity Factor (CCF) Reveals Stranded Capacity and Data Center Cost Savings

WHITE PAPER Cooling Capacity Factor (CCF) Reveals Stranded Capacity and Data Center Cost Savings By Lars Strong, P.E., Upsite Technologies, Inc. Kenneth G. Brill, Upsite Technologies, Inc. 505.798.0200

### Dynamic Thermal Management for Distributed Systems

Dynamic Thermal Management for Distributed Systems Andreas Weissel Frank Bellosa Department of Computer Science 4 (Operating Systems) University of Erlangen-Nuremberg Martensstr. 1, 91058 Erlangen, Germany

### SIDN Server Measurements

SIDN Server Measurements Yuri Schaeffer 1, NLnet Labs NLnet Labs document 2010-003 July 19, 2010 1 Introduction For future capacity planning SIDN would like to have an insight on the required resources

### Power efficiency and power management in HP ProLiant servers

Power efficiency and power management in HP ProLiant servers Technology brief Introduction... 2 Built-in power efficiencies in ProLiant servers... 2 Optimizing internal cooling and fan power with Sea of

### DataCenter 2020: hot aisle and cold aisle containment efficiencies reveal no significant differences

DataCenter 2020: hot aisle and cold aisle containment efficiencies reveal no significant differences November 2011 Powered by DataCenter 2020: hot aisle and cold aisle containment efficiencies reveal no

### Introducing the Singlechip Cloud Computer

Introducing the Singlechip Cloud Computer Exploring the Future of Many-core Processors White Paper Intel Labs Jim Held Intel Fellow, Intel Labs Director, Tera-scale Computing Research Sean Koehl Technology

### Using VMware VMotion with Oracle Database and EMC CLARiiON Storage Systems

Using VMware VMotion with Oracle Database and EMC CLARiiON Storage Systems Applied Technology Abstract By migrating VMware virtual machines from one physical environment to another, VMware VMotion can

### Agenda. Context. System Power Management Issues. Power Capping Overview. Power capping participants. Recommendations

Power Capping Linux Agenda Context System Power Management Issues Power Capping Overview Power capping participants Recommendations Introduction of Linux Power Capping Framework 2 Power Hungry World Worldwide,

### Thermal Modeling Methodology for Fast and Accurate System-Level Analysis: Application to a Memory-on-Logic 3D Circuit

Thermal Modeling Methodology for Fast and Accurate System-Level Analysis: Application to a Memory-on-Logic 3D Circuit Cristiano Santos 1,2, Pascal Vivet 1, Philippe Garrault 3, Nicolas Peltier 3, Sylvian

### Overlapping Data Transfer With Application Execution on Clusters

Overlapping Data Transfer With Application Execution on Clusters Karen L. Reid and Michael Stumm reid@cs.toronto.edu stumm@eecg.toronto.edu Department of Computer Science Department of Electrical and Computer

### Using CFD for optimal thermal management and cooling design in data centers

www.siemens.com/datacenters Using CFD for optimal thermal management and cooling design in data centers Introduction As the power density of IT equipment within a rack increases and energy costs rise,

### ConSil: Low-cost Thermal Mapping of Data Centers

ConSil: Low-cost Thermal Mapping of Data Centers Justin Moore, Jeffrey S. Chase, and Parthasarathy Ranganathan Duke University Hewlett Packard Labs {justin,chase}@cs.duke.edu partha.ranganathan@hp.com

### Introduction 1 Performance on Hosted Server 1. Benchmarks 2. System Requirements 7 Load Balancing 7

Introduction 1 Performance on Hosted Server 1 Figure 1: Real World Performance 1 Benchmarks 2 System configuration used for benchmarks 2 Figure 2a: New tickets per minute on E5440 processors 3 Figure 2b:

### Intel RAID Controllers

Intel RAID Controllers Best Practices White Paper April, 2008 Enterprise Platforms and Services Division - Marketing Revision History Date Revision Number April, 2008 1.0 Initial release. Modifications

### Understanding Linux on z/vm Steal Time

Understanding Linux on z/vm Steal Time June 2014 Rob van der Heij rvdheij@velocitysoftware.com Summary Ever since Linux distributions started to report steal time in various tools, it has been causing

### Data Center Cooling: Fend Off The Phantom Meltdown Of Mass Destruction. 670 Deer Road n Cherry Hill, NJ 08034 n 877.429.7225 n

Data Center Cooling: Fend Off The Phantom Meltdown Of Mass Destruction How To Preserve Your Servers And Prevent Overheating Your high-performance, multiprocessor servers are working hard, computing tons

Copyright www.agileload.com 1 INTRODUCTION Performance testing is a complex activity where dozens of factors contribute to its success and effective usage of all those factors is necessary to get the accurate

### - An Essential Building Block for Stable and Reliable Compute Clusters

Ferdinand Geier ParTec Cluster Competence Center GmbH, V. 1.4, March 2005 Cluster Middleware - An Essential Building Block for Stable and Reliable Compute Clusters Contents: Compute Clusters a Real Alternative

### Load Distribution in Large Scale Network Monitoring Infrastructures

Load Distribution in Large Scale Network Monitoring Infrastructures Josep Sanjuàs-Cuxart, Pere Barlet-Ros, Gianluca Iannaccone, and Josep Solé-Pareta Universitat Politècnica de Catalunya (UPC) {jsanjuas,pbarlet,pareta}@ac.upc.edu

### Performance Prediction, Sizing and Capacity Planning for Distributed E-Commerce Applications

Performance Prediction, Sizing and Capacity Planning for Distributed E-Commerce Applications by Samuel D. Kounev (skounev@ito.tu-darmstadt.de) Information Technology Transfer Office Abstract Modern e-commerce

### Optimizing Shared Resource Contention in HPC Clusters

Optimizing Shared Resource Contention in HPC Clusters Sergey Blagodurov Simon Fraser University Alexandra Fedorova Simon Fraser University Abstract Contention for shared resources in HPC clusters occurs

### IaaS Cloud Architectures: Virtualized Data Centers to Federated Cloud Infrastructures

IaaS Cloud Architectures: Virtualized Data Centers to Federated Cloud Infrastructures Dr. Sanjay P. Ahuja, Ph.D. 2010-14 FIS Distinguished Professor of Computer Science School of Computing, UNF Introduction

### Effect of Rack Server Population on Temperatures in Data Centers

Effect of Rack Server Population on Temperatures in Data Centers Rajat Ghosh, Vikneshan Sundaralingam, Yogendra Joshi G.W. Woodruff School of Mechanical Engineering Georgia Institute of Technology, Atlanta,

### Is Your Type Approved Computer Really Seaworthy?

Is Your Type Approved Computer Really Seaworthy? Introduction As more critical shipboard functions such as navigation are being computerized, performance, reliability and integrity of these systems are

### Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging

Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging In some markets and scenarios where competitive advantage is all about speed, speed is measured in micro- and even nano-seconds.

### High Performance Cluster Support for NLB on Window

High Performance Cluster Support for NLB on Window [1]Arvind Rathi, [2] Kirti, [3] Neelam [1]M.Tech Student, Department of CSE, GITM, Gurgaon Haryana (India) arvindrathi88@gmail.com [2]Asst. Professor,

### Best Practices for Optimizing Your Linux VPS and Cloud Server Infrastructure

Best Practices for Optimizing Your Linux VPS and Cloud Server Infrastructure Q1 2012 Maximizing Revenue per Server with Parallels Containers for Linux www.parallels.com Table of Contents Overview... 3

### Absolute and relative humidity Precise and comfort air-conditioning

Absolute and relative humidity Precise and comfort air-conditioning Trends and best practices in data centers Compiled by: CONTEG Date: 30. 3. 2010 Version: 1.11 EN 2013 CONTEG. All rights reserved. No

### TECHNICAL WHITE PAPER. Monitoring Cisco Hardware with Sentry Software Monitoring for BMC ProactiveNet Performance Management

TECHNICAL WHITE PAPER Monitoring Cisco Hardware with Sentry Software Monitoring for BMC ProactiveNet Performance Management Table of Contents Situation overview 1 Monitoring Cisco UCS Series 1 Monitoring

### Tableau Server 7.0 scalability

Tableau Server 7.0 scalability February 2012 p2 Executive summary In January 2012, we performed scalability tests on Tableau Server to help our customers plan for large deployments. We tested three different

### Benchmarking Hadoop & HBase on Violin

Technical White Paper Report Technical Report Benchmarking Hadoop & HBase on Violin Harnessing Big Data Analytics at the Speed of Memory Version 1.0 Abstract The purpose of benchmarking is to show advantages

### An Oracle White Paper November 2010. Oracle Real Application Clusters One Node: The Always On Single-Instance Database

An Oracle White Paper November 2010 Oracle Real Application Clusters One Node: The Always On Single-Instance Database Executive Summary... 1 Oracle Real Application Clusters One Node Overview... 1 Always

### 11.1 inspectit. 11.1. inspectit

11.1. inspectit Figure 11.1. Overview on the inspectit components [Siegl and Bouillet 2011] 11.1 inspectit The inspectit monitoring tool (website: http://www.inspectit.eu/) has been developed by NovaTec.

### Comparison of Windows IaaS Environments

Comparison of Windows IaaS Environments Comparison of Amazon Web Services, Expedient, Microsoft, and Rackspace Public Clouds January 5, 215 TABLE OF CONTENTS Executive Summary 2 vcpu Performance Summary

### White Paper on Consolidation Ratios for VDI implementations

White Paper on Consolidation Ratios for VDI implementations Executive Summary TecDem have produced this white paper on consolidation ratios to back up the return on investment calculations and savings

### Data Center Operating Temperature: The Sweet Spot

Data Center Operating Temperature: The Sweet Spot A Dell Technical White Paper Dell Data Center Infrastructure David L. Moss THIS WHITE PAPER IS FOR INFORMATIONAL PURPOSES ONLY, AND MAY CONTAIN TYPOGRAPHICAL

### TPC-W * : Benchmarking An Ecommerce Solution By Wayne D. Smith, Intel Corporation Revision 1.2

TPC-W * : Benchmarking An Ecommerce Solution By Wayne D. Smith, Intel Corporation Revision 1.2 1 INTRODUCTION How does one determine server performance and price/performance for an Internet commerce, Ecommerce,

### APC APPLICATION NOTE #112

#112 Best Practices for Deploying the InfraStruXure InRow SC By David Roden Abstract The InfraStruXure InRow SC (ACSC100 and ACSC101) is a self-contained air conditioner for server rooms and wiring closets.

### Efficient Data Replication Scheme based on Hadoop Distributed File System

, pp. 177-186 http://dx.doi.org/10.14257/ijseia.2015.9.12.16 Efficient Data Replication Scheme based on Hadoop Distributed File System Jungha Lee 1, Jaehwa Chung 2 and Daewon Lee 3* 1 Division of Supercomputing,

### The Mainframe Virtualization Advantage: How to Save Over Million Dollars Using an IBM System z as a Linux Cloud Server

Research Report The Mainframe Virtualization Advantage: How to Save Over Million Dollars Using an IBM System z as a Linux Cloud Server Executive Summary Information technology (IT) executives should be

### Data centers have been cooled for many years by delivering

The Increasing Challenge of Data Center Design and Management: Is CFD a Must? Mark Seymour, Christopher Aldham, Matthew Warner, Hassan Moezzi Future Facilities, San Jose, California, USA Mark Seymour,

### A Taxonomy and Survey of Energy-Efficient Data Centers and Cloud Computing Systems

A Taxonomy and Survey of Energy-Efficient Data Centers and Cloud Computing Systems Anton Beloglazov, Rajkumar Buyya, Young Choon Lee, and Albert Zomaya Present by Leping Wang 1/25/2012 Outline Background

### Microsoft Windows Server 2003 with Internet Information Services (IIS) 6.0 vs. Linux Competitive Web Server Performance Comparison

April 23 11 Aviation Parkway, Suite 4 Morrisville, NC 2756 919-38-28 Fax 919-38-2899 32 B Lakeside Drive Foster City, CA 9444 65-513-8 Fax 65-513-899 www.veritest.com info@veritest.com Microsoft Windows

### Why Computers Are Getting Slower (and what we can do about it) Rik van Riel Sr. Software Engineer, Red Hat

Why Computers Are Getting Slower (and what we can do about it) Rik van Riel Sr. Software Engineer, Red Hat Why Computers Are Getting Slower The traditional approach better performance Why computers are

### Thermal Modeling of NiMH Battery Powering Electric Vehicles

Proceedings of the 5th WSEAS Int. Conf. on System Science and Simulation in Engineering, Tenerife, Canary Islands, Spain, December 16-18, 06 5 Thermal ing of NiMH Battery Powering Electric Vehicles H.

### White Paper Perceived Performance Tuning a system for what really matters

TMurgent Technologies White Paper Perceived Performance Tuning a system for what really matters September 18, 2003 White Paper: Perceived Performance 1/7 TMurgent Technologies Introduction The purpose

### Introducing AUDIT-BUDDY

Introducing AUDIT-BUDDY Optimize the Data Center with AUDIT-BUDDY Executive Summary Proper temperature and humidity for the inlet air into the servers is essential to efficient data center operation. Several

### InfiniteGraph: The Distributed Graph Database

A Performance and Distributed Performance Benchmark of InfiniteGraph and a Leading Open Source Graph Database Using Synthetic Data Objectivity, Inc. 640 West California Ave. Suite 240 Sunnyvale, CA 94086

### Boosting the Energy efficiency of a Server Room: experimental + virtual approach.

Boosting the Energy efficiency of a Server Room: experimental + virtual approach. Nowadays, the increase in computing power allows the successful applications of the CFD simulation methods in civil environments.

### Relational Databases in the Cloud

Contact Information: February 2011 zimory scale White Paper Relational Databases in the Cloud Target audience CIO/CTOs/Architects with medium to large IT installations looking to reduce IT costs by creating

### HyperThreading Support in VMware ESX Server 2.1

HyperThreading Support in VMware ESX Server 2.1 Summary VMware ESX Server 2.1 now fully supports Intel s new Hyper-Threading Technology (HT). This paper explains the changes that an administrator can expect

### INTERNATIONAL JOURNAL OF MECHANICAL ENGINEERING AND TECHNOLOGY (IJMET)

INTERNATIONAL JOURNAL OF MECHANICAL ENGINEERING AND TECHNOLOGY (IJMET) Proceedings of the 2 nd International Conference on Current Trends in Engineering and Management ICCTEM -2014 ISSN 0976 6340 (Print)

### White Paper. Real-time Capabilities for Linux SGI REACT Real-Time for Linux

White Paper Real-time Capabilities for Linux SGI REACT Real-Time for Linux Abstract This white paper describes the real-time capabilities provided by SGI REACT Real-Time for Linux. software. REACT enables

### Oracle Database Scalability in VMware ESX VMware ESX 3.5

Performance Study Oracle Database Scalability in VMware ESX VMware ESX 3.5 Database applications running on individual physical servers represent a large consolidation opportunity. However enterprises

### III Big Data Technologies

III Big Data Technologies Today, new technologies make it possible to realize value from Big Data. Big data technologies can replace highly customized, expensive legacy systems with a standard solution

### Energy Efficient MapReduce

Energy Efficient MapReduce Motivation: Energy consumption is an important aspect of datacenters efficiency, the total power consumption in the united states has doubled from 2000 to 2005, representing

### What Is Specific in Load Testing?

What Is Specific in Load Testing? Testing of multi-user applications under realistic and stress loads is really the only way to ensure appropriate performance and reliability in production. Load testing

### Icepak High-Performance Computing at Rockwell Automation: Benefits and Benchmarks

Icepak High-Performance Computing at Rockwell Automation: Benefits and Benchmarks Garron K. Morris Senior Project Thermal Engineer gkmorris@ra.rockwell.com Standard Drives Division Bruce W. Weiss Principal

### 1. Comments on reviews a. Need to avoid just summarizing web page asks you for:

1. Comments on reviews a. Need to avoid just summarizing web page asks you for: i. A one or two sentence summary of the paper ii. A description of the problem they were trying to solve iii. A summary of

### Improving Rack Cooling Performance Using Airflow Management Blanking Panels

Improving Rack Cooling Performance Using Airflow Management Blanking Panels By Neil Rasmussen White Paper #44 Revision 3 Executive Summary Unused vertical space in open frame racks and rack enclosures

### The CEETHERM Data Center Laboratory

The CEETHERM Data Center Laboratory A Platform for Transformative Research on Green Data Centers Yogendra Joshi and Pramod Kumar G.W. Woodruff School of Mechanical Engineering Georgia Institute of Technology

### Power Consumption and Cooling in the Data Center: A Survey

A M a r k e t A n a l y s i s Power Consumption and Cooling in the Data Center: A Survey 1 Executive Summary Power consumption and cooling issues in the data center are important to companies, as evidenced

### Benchmarking Cassandra on Violin

Technical White Paper Report Technical Report Benchmarking Cassandra on Violin Accelerating Cassandra Performance and Reducing Read Latency With Violin Memory Flash-based Storage Arrays Version 1.0 Abstract

### Power and Cooling Innovations in Dell PowerEdge Servers

Power and Cooling Innovations in Dell PowerEdge Servers This technical white paper describes the Dell Energy Smart Architecture and the new and enhanced features designed into Dell PowerEdge 12 th generation

### Verizon SMARTS Data Center Design Phase 1 Conceptual Study Report Ms. Leah Zabarenko Verizon Business 2606A Carsins Run Road Aberdeen, MD 21001

Verizon SMARTS Data Center Design Phase 1 Conceptual Study Report Ms. Leah Zabarenko Verizon Business 2606A Carsins Run Road Aberdeen, MD 21001 Presented by: Liberty Engineering, LLP 1609 Connecticut Avenue

### 10.04.2008. Thomas Fahrig Senior Developer Hypervisor Team. Hypervisor Architecture Terminology Goals Basics Details

Thomas Fahrig Senior Developer Hypervisor Team Hypervisor Architecture Terminology Goals Basics Details Scheduling Interval External Interrupt Handling Reserves, Weights and Caps Context Switch Waiting

### ENERGY STAR Data Center Storage Version 1.0. Stakeholder Update. 02 February 2010

ENERGY STAR Data Center Storage Version 1.0 Stakeholder Update 02 February 2010 Agenda Development Schedule Overview Draft 1 Considerations Partner Commitments Definitions Product Family Taxonomy PSU Criteria

### Power and Cooling for Ultra-High Density Racks and Blade Servers

Power and Cooling for Ultra-High Density Racks and Blade Servers White Paper #46 Introduction The Problem Average rack in a typical data center is under 2 kw Dense deployment of blade servers (10-20 kw

### Unique Airflow Visualization Techniques for the Design and Validation of Above-Plenum Data Center CFD Models

Unique Airflow Visualization Techniques for the Design and Validation of Above-Plenum Data Center CFD Models The MIT Faculty has made this article openly available. Please share how this access benefits

### CoolEmAll - Tools for realising an energy efficient data centre

CoolEmAll - Tools for realising an energy efficient data centre Wolfgang Christmann christmann informationstechnik + medien GmbH & Co. KG www.christmann.info 1 Outline CoolEmAll project RECS system towards

### Abstract: Motivation: Description of proposal:

Efficient power utilization of a cluster using scheduler queues Kalyana Chadalvada, Shivaraj Nidoni, Toby Sebastian HPCC, Global Solutions Engineering Bangalore Development Centre, DELL Inc. {kalyana_chadalavada;shivaraj_nidoni;toby_sebastian}@dell.com

### Reducing Data Center Energy Consumption

Reducing Data Center Energy Consumption By John Judge, Member ASHRAE; Jack Pouchet, Anand Ekbote, and Sachin Dixit Rising data center energy consumption and increasing energy costs have combined to elevate

### The following are general terms that we have found being used by tenants, landlords, IT Staff and consultants when discussing facility space.

The following are general terms that we have found being used by tenants, landlords, IT Staff and consultants when discussing facility space. Terminology: Telco: Dmarc: NOC: SAN: GENSET: Switch: Blade

### Server Consolidation with SQL Server 2008

Server Consolidation with SQL Server 2008 White Paper Published: August 2007 Updated: July 2008 Summary: Microsoft SQL Server 2008 supports multiple options for server consolidation, providing organizations

### An Implementation Of Multiprocessor Linux

An Implementation Of Multiprocessor Linux This document describes the implementation of a simple SMP Linux kernel extension and how to use this to develop SMP Linux kernels for architectures other than

### THERMAL DESIGN AND TEST REQUIREMENTS FOR OUTSIDE PLANT CABLE TELECOMMUNICATIONS EQUIPMENT Al Marshall, P.E. Philips Broadband Networks

THERMAL DESIGN AND TEST REQUIREMENTS FOR OUTSIDE PLANT CABLE TELECOMMUNICATIONS EQUIPMENT Al Marshall, P.E. Philips Broadband Networks Abstract Shrinking thermal margins, driven by sophisticated but thermally

### THE CFD SIMULATION OF THE FLOW AROUND THE AIRCRAFT USING OPENFOAM AND ANSA

THE CFD SIMULATION OF THE FLOW AROUND THE AIRCRAFT USING OPENFOAM AND ANSA Adam Kosík Evektor s.r.o., Czech Republic KEYWORDS CFD simulation, mesh generation, OpenFOAM, ANSA ABSTRACT In this paper we describe

OpenFlow Based Load Balancing Hardeep Uppal and Dane Brandon University of Washington CSE561: Networking Project Report Abstract: In today s high-traffic internet, it is often desirable to have multiple

### pc resource monitoring and performance advisor

pc resource monitoring and performance advisor application note www.hp.com/go/desktops Overview HP Toptools is a modular web-based device management tool that provides dynamic information about HP hardware