Datacenter as a Computer(Edition 2).pdf

发布时间：2022-06-05 发布人：admin 分类：说明书资料大小：2.93M 资料格式：pdf 举报版权申诉

laynepeng-9532683-4744302543315892194.pdf-第1页.png

第1页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第2页.png

第2页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第3页.png

第3页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第4页.png

第4页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第5页.png

第5页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第6页.png

第6页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第7页.png

第7页 / 共156页

laynepeng-9532683-4744302543315892194.pdf-第8页.png

第8页 / 共156页

Acknowledgments

Note to the Reader

Introduction

1.1 Warehouse-Scale Computers

1.2 Cost Efficiency at Scale

1.3 Not Just a Collection of Servers

1.4 One Datacenter Vs. Several Datacenters

1.5 Why WSCS Might Matter to You

1.6 Architectural Overview of WSCS

1.6.1 Storage

1.6.2 Networking Fabric

1.6.3 Storage Hierarchy

1.6.4 Quantifying Latency, Bandwidth, and Capacity

1.6.5 Power Usage

1.6.6 Handling Failures

Workloads and Software Infrastructure

2.1 Datacenter Vs. Desktop

2.2 Performance and Availability Toolbox

2.3 Platform-Level Software

2.4 Cluster-Level Infrastructure Software

2.4.1 Resource Management

2.4.2 Hardware Abstraction and Other Basic Services

2.4.3 Deployment and Maintenance

2.4.4 Programming Frameworks

2.5 Application-Level Software

2.5.1 Workload Examples

2.5.2 Online: Web Search

2.5.3 Offline: Scholar Article Similarity

2.6 A Monitoring Infrastructure

2.6.1 Service-Level Dashboards

2.6.2 Performance Debugging Tools

2.6.3 Platform-Level Health Monitoring

2.7 Buy Vs. Build

2.8 Tail-Tolerance

2.9 Further Reading

Hardware Building Blocks

3.1 Cost-Efficient Server Hardware

3.1.1 The Impact of Large SMP Communication Efficiency

3.1.2 Brawny vs. Wimpy Servers

3.1.3 Balanced Designs

3.2 WSC Storage

3.2.1 Unstructured WSC Storage

3.2.2 Structured WSC Storage

3.2.3 Interplay of Storage and Networking Technology

3.3 WSC Networking

3.4 Further Reading

Datacenter Basics

4.1 Datacenter Tier Classifications and Specifications

4.2 Datacenter Power Systems

4.2.1 Uninterruptible Power Systems

4.2.2 Power Distribution Units

4.2.3 Alternative: DC Distribution

4.3 Datacenter Cooling Systems

4.3.1 CRACs, Chillers, and Cooling Towers

4.3.2 CRACs

4.3.3 Chillers

4.3.4 Cooling towers

4.3.5 Free Cooling

4.3.6 Air Flow Considerations

4.3.7 In-Rack, In-Row Cooling, and Cold Plates

4.3.8 Case Study: Google’s In-row Cooling

4.3.9 Container-Based Datacenters

4.4 Summary

Energy and Power Efficiency

5.1 Datacenter Energy Efficiency

5.1.1 The PUE Metric

5.1.2 Issues with the PUE Metric

5.1.3 Sources of Efficiency Losses in Datacenters

5.1.4 Improving the Energy Efficiency of Datacenters

5.1.5 Beyond the Facility

5.2 The Energy Efficiency of Computing

5.2.1 Measuring Energy Efficiency

5.2.2 Server Energy Efficiency

5.2.3 Usage Profile of Warehouse-Scale Computers

5.3 Energy-Proportional Computing

5.3.1 Causes of Poor Energy Proportionality

5.3.2 Improving Energy Proportionality

5.3.3 Energy Proportionality—The Rest of the System

5.4 Relative Effectiveness of Low-Power Modes

5.5 The Role of Software in Energy Proportionality

5.6 Datacenter Power Provisioning

5.6.1 Deploying the Right Amount of Equipment

5.6.2 Oversubscribing Facility Power

5.7 Trends in Server Energy Usage

5.7.1 Using Energy Storage for Power Management

5.8 Conclusions

5.8.1 Further Reading

Modeling Costs

6.1 Capital Costs

6.2 Operational Costs

6.3 Case Studies

6.3.1 Real-World Datacenter Costs

6.3.2 Modeling a Partially Filled Datacenter

6.3.3 The Cost of Public Clouds

Dealing with Failures and Repairs

7.1 Implications of Software-Based Fault Tolerance

7.2 Categorizing Faults

7.3 Machine-Level Failures

7.4 Repairs

7.5 Tolerating Faults, Not Hiding Them

Closing Remarks

8.1 Hardware

8.2 Software

8.3 Economics

8.4 Key Challenges

8.4.1 Rapidly Changing Workloads

8.4.2 Building Responsive Large Scale Systems

8.4.3 Energy Proportionality of Non-CPU components

8.4.4 Overcoming the End of Dennard Scaling

8.4.5 Amdahl’s Cruel Law

8.5 Conclusions

Bibliography

Author Biographies

Series ISSN: 1935-3235 Series ISSN: 1935-3235 Series ISSN: 1935-3235 SYNTHESIS LECTURES ON SYNTHESIS LECTURES ON SYNTHESIS LECTURES ON COMPUTER ARCHITECTURE COMPUTER ARCHITECTURE COMPUTER ARCHITECTURE Series Editor: Mark D. Hill, University of Wisconsin Series Editor: Mark D. Hill, University of Wisconsin Series Editor: Mark D. Hill, University of Wisconsin An Introduction to the Design of Warehouse-Scale Machines, Second Edition An Introduction to the Design of Warehouse-Scale Machines, Second Edition An Introduction to the Design of Warehouse-Scale Machines, Second Edition Google, Inc. Google, Inc. Google, Inc. As computation continues to move into the cloud, the computing platform of interest no longer resembles a As computation continues to move into the cloud, the computing platform of interest no longer resembles a As computation continues to move into the cloud, the computing platform of interest no longer resembles a pizza box or a refrigerator, but a warehouse full of computers. These new large datacenters are quite different pizza box or a refrigerator, but a warehouse full of computers. These new large datacenters are quite different pizza box or a refrigerator, but a warehouse full of computers. These new large datacenters are quite different from traditional hosting facilities of earlier times and cannot be viewed simply as a collection of co-located from traditional hosting facilities of earlier times and cannot be viewed simply as a collection of co-located from traditional hosting facilities of earlier times and cannot be viewed simply as a collection of co-located servers. Large portions of the hardware and software resources in these facilities must work in concert to servers. Large portions of the hardware and software resources in these facilities must work in concert to servers. Large portions of the hardware and software resources in these facilities must work in concert to efficiently deliver good levels of Internet service performance, something that can only be achieved by a holistic efficiently deliver good levels of Internet service performance, something that can only be achieved by a holistic efficiently deliver good levels of Internet service performance, something that can only be achieved by a holistic approach to their design and deployment. In other words, we must treat the datacenter itself as one massive approach to their design and deployment. In other words, we must treat the datacenter itself as one massive approach to their design and deployment. In other words, we must treat the datacenter itself as one massive warehouse-scale computer (WSC). We describe the architecture of WSCs, the main factors influencing their warehouse-scale computer (WSC). We describe the architecture of WSCs, the main factors influencing their warehouse-scale computer (WSC). We describe the architecture of WSCs, the main factors influencing their design, operation, and cost structure, and the characteristics of their software base. We hope it will be useful design, operation, and cost structure, and the characteristics of their software base. We hope it will be useful design, operation, and cost structure, and the characteristics of their software base. We hope it will be useful to architects and programmers of today’s WSCs, as well as those of future many-core platforms which may to architects and programmers of today’s WSCs, as well as those of future many-core platforms which may to architects and programmers of today’s WSCs, as well as those of future many-core platforms which may one day implement the equivalent of today’s WSCs on a single board. one day implement the equivalent of today’s WSCs on a single board. one day implement the equivalent of today’s WSCs on a single board. After nearly four years of substantial academic and industrial developments in warehouse-scale computing, After nearly four years of substantial academic and industrial developments in warehouse-scale computing, After nearly four years of substantial academic and industrial developments in warehouse-scale computing, we are delighted to present our first major update to this lecture. The increased popularity of public clouds we are delighted to present our first major update to this lecture. The increased popularity of public clouds we are delighted to present our first major update to this lecture. The increased popularity of public clouds has made WSC software techniques relevant to a larger pool of programmers since our first edition. Therefore, has made WSC software techniques relevant to a larger pool of programmers since our first edition. Therefore, has made WSC software techniques relevant to a larger pool of programmers since our first edition. Therefore, we expanded Chapter 2 to reflect our better understanding of WSC software systems and the toolbox of we expanded Chapter 2 to reflect our better understanding of WSC software systems and the toolbox of we expanded Chapter 2 to reflect our better understanding of WSC software systems and the toolbox of software techniques for WSC programming. In Chapter 3, we added to our coverage of the evolving landscape software techniques for WSC programming. In Chapter 3, we added to our coverage of the evolving landscape software techniques for WSC programming. In Chapter 3, we added to our coverage of the evolving landscape of wimpy vs. brawny server trade-offs, and we now present an overview of WSC interconnects and storage of wimpy vs. brawny server trade-offs, and we now present an overview of WSC interconnects and storage of wimpy vs. brawny server trade-offs, and we now present an overview of WSC interconnects and storage systems that was promised but lacking in the original edition. Thanks largely to the help of our new co-author, systems that was promised but lacking in the original edition. Thanks largely to the help of our new co-author, systems that was promised but lacking in the original edition. Thanks largely to the help of our new co-author, Google Distinguished Engineer Jimmy Clidaras, the material on facility mechanical and power distribution Google Distinguished Engineer Jimmy Clidaras, the material on facility mechanical and power distribution Google Distinguished Engineer Jimmy Clidaras, the material on facility mechanical and power distribution design has been updated and greatly extended (see Chapters 4 and 5). Chapters 6 and 7 have also been design has been updated and greatly extended (see Chapters 4 and 5). Chapters 6 and 7 have also been design has been updated and greatly extended (see Chapters 4 and 5). Chapters 6 and 7 have also been revamped significantly. We hope this revised edition continues to meet the needs of educators and professionals revamped significantly. We hope this revised edition continues to meet the needs of educators and professionals revamped significantly. We hope this revised edition continues to meet the needs of educators and professionals in this area. in this area. in this area. ABOUT SYNTHESIS ABOUT SYNTHESIS ABOUT SYNTHESIS This volume is a printed version of a work that appears in the Synthesis This volume is a printed version of a work that appears in the Synthesis This volume is a printed version of a work that appears in the Synthesis Digital Library of Engineering and Computer Science. Synthesis Lectures Digital Library of Engineering and Computer Science. Synthesis Lectures Digital Library of Engineering and Computer Science. Synthesis Lectures provide concise, original presentations of important research and development provide concise, original presentations of important research and development provide concise, original presentations of important research and development topics, published quickly, in digital and print formats. For more information topics, published quickly, in digital and print formats. For more information topics, published quickly, in digital and print formats. For more information visit www.morganclaypool.com visit www.morganclaypool.com visit www.morganclaypool.com & & & Morgan Claypool Publishers Morgan Claypool Publishers Morgan Claypool Publishers ISBN: 978-1-62705-009-8 ISBN: 978-1-62705-009-8 ISBN: 978-1-62705-009-8 90000 90000 90000 w w w . m o r g a n c l a y p o o l . c o m w w w . m o r g a n c l a y p o o l . c o m w w w . m o r g a n c l a y p o o l . c o m 9 781627 050098 9 781627 050098 9 781627 050098 & & & CM& Morgan Claypool Publishers CM& Morgan Claypool Publishers CM& Morgan Claypool Publishers An Introduction to the Design An Introduction to the Design An Introduction to the Design of Warehouse-Scale Machines of Warehouse-Scale Machines of Warehouse-Scale Machines Second Edition Second Edition Second Edition M M M O O O R R R G G G A A A N N N & & & C C C L L L A A A Y Y Y P P P O O O O O O L L L SYNTHESIS LECTURES ON SYNTHESIS LECTURES ON SYNTHESIS LECTURES ON COMPUTER ARCHITECTURE COMPUTER ARCHITECTURE COMPUTER ARCHITECTURE Mark D. Hill, Series Editor Mark D. Hill, Series Editor Mark D. Hill, Series Editor

!e Datacenter as a Computer An Introduction to the Design of Warehouse-Scale Machines Second Edition

iii Synthesis Lectures on Computer Architecture Editor Mark D. Hill, University of Wisconsin, Madison Synthesis Lectures on Computer Architecture publishes 50- to 100-page publications on topics pertaining to the science and art of designing, analyzing, selecting, and interconnecting hardware components to create computers that meet functional, performance, and cost goals. !e scope will largely follow the purview of premier computer architecture conferences, such as ISCA, HPCA, MICRO, and ASPLOS. !e Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines, Second Edition Luiz André Barroso, Jimmy Clidaras, and Urs Hölzle 2013 Shared-Memory Synchronization Michael L. Scott 2013 Resilient Architecture Design for Voltage Variation Vijay Janapa Reddi , Meeta Sharma Gupta 2013 Multithreading Architecture Mario Nemirovsky, Dean M. Tullsen 2013 Performance Analysis and Tuning for General Purpose Graphics Processing Units (GPGPU) Hyesoon Kim, Richard Vuduc, Sara Baghsorkhi, Jee Choi, Wen-mei Hwu 2012 Automatic Parallelization: An Overview of Fundamental Compiler Techniques Samuel P. Midki" 2012

iv Phase Change Memory: From Devices to Systems Moinuddin K. Qureshi, Sudhanva Gurumurthi, Bipin Rajendran 2011 Multi-Core Cache Hierarchies Rajeev Balasubramonian, Norman P. Jouppi, Naveen Muralimanohar 2011 A Primer on Memory Consistency and Cache Coherence Daniel J. Sorin, Mark D. Hill, David A. Wood 2011 Dynamic Binary Modi"cation: Tools, Techniques, and Applications Kim Hazelwood 2011 Quantum Computing for Computer Architects, Second Edition Tzvetan S. Metodi, Arvin I. Faruque, Frederic T. Chong 2011 High Performance Datacenter Networks: Architectures, Algorithms, and Opportunities Dennis Abts, John Kim 2011 Processor Microarchitecture: An Implementation Perspective Antonio González, Fernando Latorre, Grigorios Magklis 2010 Transactional Memory, 2nd edition Tim Harris , James Larus , Ravi Rajwar 2010 Computer Architecture Performance Evaluation Methods Lieven Eeckhout 2010 Introduction to Recon"gurable Supercomputing Marco Lanzagorta, Stephen Bique, Robert Rosenberg 2009 On-Chip Networks Natalie Enright Jerger, Li-Shiuan Peh 2009

v !e Memory System: You Can’t Avoid It, You Can’t Ignore It, You Can’t Fake It Bruce Jacob 2009 Fault Tolerant Computer Architecture Daniel J. Sorin 2009 !e Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines Luiz André Barroso, Urs Hölzle 2009 Computer Architecture Techniques for Power-E#ciency Stefanos Kaxiras, Margaret Martonosi 2008 Chip Multiprocessor Architecture: Techniques to Improve !roughput and Latency Kunle Olukotun, Lance Hammond, James Laudon 2007 Transactional Memory James R. Larus, Ravi Rajwar 2006 Quantum Computing for Computer Architects Tzvetan S. Metodi, Frederic T. Chong 2006

Copyright © 2013 by Morgan & Claypool All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means—electronic, mechanical, photocopy, recording, or any other except for brief quotations in printed reviews, without the prior permission of the publisher. !e Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines, Second Edition Luiz André Barroso, Jimmy Clidaras, and Urs Hölzle www.morganclaypool.com ISBN: 9781627050098 print ISBN: 9781627050104 ebook DOI 10.2200/S00516ED2V01Y201306CAC024 A Publication in the Morgan & Claypool Publishers series SYNTHESIS LECTURES ON COMPUTER ARCHITECTURE Lecture #24 Series Editor: Mark D. Hill, University of Wisconsin, Madison Series ISSN 1935-3235 Print 1935-3243 Electronic

分享到：

赞收藏

资料库

Datacenter as a Computer(Edition 2).pdf

相关推荐

服务器应用

热门标签

最新资料