Cloud Data Centers and Cost Modeling (eBook)

A Complete Guide To Planning, Designing and Building a Cloud Data Center

Rajkumar Buyya, Caesar Wu (Autoren)

eBook Download: EPUB

2015 | 1. Auflage
848 Seiten
Elsevier Science (Verlag)
978-0-12-801688-6 (ISBN)

Infos

Cloud Data Centers and Cost Modeling establishes a framework for strategic decision-makers to facilitate the development of cloud data centers. Just as building a house requires a clear understanding of the blueprints, architecture, and costs of the project; building a cloud-based data center requires similar knowledge. The authors take a theoretical and practical approach, starting with the key questions to help uncover needs and clarify project scope. They then demonstrate probability tools to test and support decisions, and provide processes that resolve key issues. After laying a foundation of cloud concepts and definitions, the book addresses data center creation, infrastructure development, cost modeling, and simulations in decision-making, each part building on the previous. In this way the authors bridge technology, management, and infrastructure as a service, in one complete guide to data centers that facilitates educated decision making. - Explains how to balance cloud computing functionality with data center efficiency - Covers key requirements for power management, cooling, server planning, virtualization, and storage management - Describes advanced methods for modeling cloud computing cost including Real Option Theory and Monte Carlo Simulations - Blends theoretical and practical discussions with insights for developers, consultants, and analysts considering data center development

Caesar Wu is a Senior Domain Specialist on Cloud Computing and Data Centers at Telstra, as well as a Principle Research Fellow, at The University of Melbourne, Australia. He has over 18 years' of experience in ICT architecture, solution design, services delivery and operation management, IT data center lifecycle and transformation. For the past five years he has been responsible for cost modeling of all Telstra cloud computing projects, for both enterprise and government clients, and designed and managed eight data centers in Australia. In 2012, Wu supervised three University of Melbourne PhD students in cloud computing strategic investment decision making.

Inhalt

Front Cover 1
Cloud Data Centers and Cost Modeling 4
Copyright Page 5
Contents 6
Preface 18
Organisation of the Book 21
Acknowledgments 22
I. Cloud Computing Foundations and Business Requirements 24
1 Cloud Computing 26
1.1 Introduction 26
1.1.1 Operation Cost Rationalization 26
1.1.2 Revenue Estimation for Emerging Products 28
1.2 Cloud Computing at a Glance 30
1.3 Right Approach to Definition 31
1.4 A Brief History of Cloud Computing Definitions 32
1.5 Parallel Computing 39
1.5.1 Hardware Parallelism 40
1.5.1.1 Processor parallelism 40
1.5.1.2 Memory parallelism 40
1.5.2 Software Parallelism 41
1.5.2.1 Algorithm parallelism 41
1.5.2.2 Programming parallelism 42
1.5.2.3 Data parallelism 42
1.5.2.4 Architecture balance parallelism 42
1.5.3 Different Types of Parallel Models 43
1.6 Distributed Computing 47
1.7 Grid Computing 48
1.8 Utility Computing 50
1.9 Cloud Computing 53
1.10 Summary 62
1.10.1 Software (Applications) 62
1.10.2 IT Infrastructure (Hardware) 63
1.11 Review Questions 64
2 Business Needs 66
2.1 Introduction 66
2.2 Project Contents and Processes 71
2.3 Allocate the Right People for the Right Job 72
2.4 Business Analyst Role 74
2.5 Defining Business 80
2.6 Business Variables 82
2.6.1 Business Entity 82
2.6.2 Business Strategy 83
2.6.3 Business Profile (Variety) 85
2.6.4 Business Size (Volume) 85
2.6.5 Business Variation 87
2.7 Classification of Business Requirements 88
2.7.1 Business Requirements 89
2.7.2 Stakeholder requirements 89
2.7.3 Solution Requirements 89
2.7.3.1 Functional requirements 91
2.7.3.2 Nonfunctional requirements 91
2.7.4 Transition Requirements 91
2.8 E2E Process of Business Problem Solving 91
2.8.1 Business Problem Definition 94
2.8.1.1 Preliminary definition 95
2.8.1.2 Analysis process 96
2.8.1.3 Confirmation and documentation of the real problem 97
2.8.1.4 Challenges of problem definition 99
2.8.1.4.1 Barking up the wrong tree 99
2.8.1.4.2 Solution side effects 99
2.8.1.4.3 Complex problems 99
2.8.1.4.4 Hidden or avoided problems 100
2.8.1.4.5 Sensitive problems 100
2.8.1.4.6 Presenting the wrong information 100
2.8.1.4.7 No single solution for the problem 100
2.8.2 Goals of Defining Business Problems 100
2.8.3 Techniques for Identifying Real Problems 101
2.8.4 Business Requirements Gathering Phase 101
2.8.4.1 Preparation 102
2.8.4.2 Conducting eliciting 103
2.8.4.3 Documenting 103
2.8.4.4 Updating 104
2.8.5 Provide the Right Solution 104
2.8.5.1 Information processing 105
2.8.5.1.1 Information classification 105
2.8.5.1.2 Information prioritization 105
2.8.5.1.3 Current process analysis 107
2.8.5.1.4 Historic event analysis 107
2.8.5.2 Modeling process 107
2.8.5.2.1 Assumptions 108
2.8.5.2.2 Data modeling 108
2.8.5.2.3 Process modeling 110
2.8.5.2.4 Behavior modeling 110
2.8.5.3 Solution process 111
2.8.5.3.1 Solution assessment 111
2.8.5.3.2 Gap analysis 111
2.8.5.3.3 Determining the best solution 112
2.8.5.3.4 Understanding the constraints of the solution 113
2.8.5.4 Communication process 113
2.8.5.4.1 Presentation and walking through the solution 113
2.8.5.4.2 Interpretation 113
2.8.5.4.3 Confirmation 114
2.8.5.4.4 Confirmation upgrading 114
2.9 Managing Expectations 114
2.10 Summary 117
2.11 Review Questions 118
3 Identifying Business Problems: A Case Study 120
3.1 Case Information Briefing 120
3.1.1 Servers 122
3.1.1.1 x86 servers: HP 122
3.1.1.2 RISC servers: Oracle/Sun E25K 122
3.1.1.2.1 E25K RISC server details 122
3.1.1.2.2 Maintenance and support requirements for E25K 124
3.1.1.2.3 Space requirements for E25K frame 126
3.1.1.2.4 Power and cooling requirements of E25K 126
3.1.1.2.5 Application requirements of E25K 129
3.1.1.3 Service contract for all RISC servers 129
3.1.2 Storage 132
3.1.2.1 NAS 132
3.1.2.2 SAN 132
3.1.3 Storage Switches 134
3.2 Define the Problems 136
3.2.1 Elicit Multiple Issues 139
3.2.1.1 Decision making 139
3.2.1.1.1 Decision motivation 140
3.2.1.1.2 Decision information and knowledge 141
3.2.1.1.3 Decision process 142
3.2.1.2 Cost transparency issue 146
3.2.1.3 Application migration issue 146
3.2.2 IT Asset Operation Practice 147
3.2.2.1 Horizontal brick wall effects 147
3.2.2.2 Vertical filtering effect 148
3.2.3 IT Operational Structure 150
3.2.3.1 Too many management layers 150
3.2.3.2 Too many IT organization changes 151
3.2.4 Misguided Incentive System 151
3.2.4.1 Wrong reason for promotion 152
3.2.4.2 IT contractors managing permanent employees 152
3.2.4.3 Salary bottleneck 153
3.3 Requirements 153
3.3.1 Business Application Requirements 155
3.3.2 Architecture Requirements 158
3.3.3 Operational Requirements 160
3.3.3.1 Shared infrastructure requirements (constraints) 160
3.3.3.2 System integration or transition requirements 160
3.3.3.3 Service monitoring requirements 160
3.3.3.4 Service maintenance and support requirements 161
3.3.4 Vendor Requirements 161
3.3.5 Other Stakeholder Requirements 164
3.3.6 Identify Hidden Requirements 164
3.4 Solution 164
3.4.1 Organizational Perspective 164
3.4.2 Technical Perspective 166
3.4.2.1 Problem statement 168
3.4.2.2 ICT’s IT strategy or business requirements 168
3.4.2.3 Assumptions 168
3.4.2.4 Proposed interim solution 168
3.4.2.5 Issues with the proposed solution 169
3.5 Summary 171
3.6 Review Questions 173
II. Data Center Facilities and Cost 174
4 Data Center Facilities 176
4.1 Basic Understanding of a Data Center 176
4.1.1 Definition of Data Center 176
4.1.2 Data Center Architecture 179
4.2 Data Center Capacity Planning 180
4.2.1 Data Center Site Selection 185
4.2.1.1 The environment 187
4.2.1.2 The power 189
4.2.1.3 The payload and IT workload 191
4.2.1.4 The policy 191
4.2.1.5 The human factor 191
4.2.1.6 The network 191
4.2.2 Data Center Performance 195
4.2.2.1 Site availability 196
4.2.2.2 Problem response and resolution time 197
4.2.2.3 Scalability 197
4.2.2.4 Utilization 198
4.2.2.5 Latency and throughput 198
4.2.3 Data Center Resource Celling 201
4.3 Data Center Space 203
4.3.1 Five Types of Space 204
4.3.1.1 Total space (building shell) 204
4.3.1.2 Total adjacent lot size (raw lot size) 206
4.3.1.3 Whitespace (raised floor) 206
4.3.1.4 Effective usable space (rack space) 207
4.3.1.5 General space 207
4.3.2 Data Center Functional Rooms 208
4.3.2.1 Utility support functions 209
4.3.2.1.1 Mechanical rooms 209
4.3.2.1.2 Electrical rooms 209
4.3.2.1.3 Staging area 209
4.3.2.2 Computing functions 210
4.3.2.2.1 Entrance rooms 210
4.3.2.2.2 Computer rooms 210
4.3.2.2.3 Telecommunication rooms 211
4.3.2.3 Operational functions 211
4.3.2.3.1 Network operation rooms 211
4.3.2.3.2 Common area 211
4.3.2.3.3 General office space 211
4.4 How to Estimate Cost of Space 212
4.5 Summary 213
4.6 Review Questions 214
5 Data Center Power 216
5.1 Introduction 216
5.2 Fundamentals of Power 218
5.2.1 Three Basic Power Metrics 218
5.2.2 Power Factor for AC Power 219
5.3 Power Panel (Circuit Breaker) 221
5.3.1 Type of Circuit Breaker and Selection 221
5.3.2 Circuit Breaker Coordination 223
5.4 Transfer Switches and Generators 223
5.4.1 Static Transfer Switch (STS) 225
5.4.2 Automatic transfer switch (ATS) 225
5.4.3 Generator 225
5.5 Uninterruptible Power Supply (UPS) 230
5.5.1 Different Types of UPS Topologies 233
5.5.1.1 Standby or offline single UPS topology 234
5.5.1.2 Line interactive UPS topology 235
5.5.1.3 Online double conversion 235
5.5.1.4 Delta conversion topology 235
5.5.1.5 Rotary UPS topology 236
5.6 How to Select UPS Topologies 236
5.6.1 UPS Redundancy and Cost Efficiency 238
5.6.1.1 Configuration of UPS redundancy 238
5.6.1.2 Single module system (SMS) 239
5.6.1.3 1+1 redundancy or two module system 239
5.6.1.4 N+1 redundancy 239
5.6.1.5 2(N+1) redundancy 241
5.6.1.6 How to balance UPS availability and cost 242
5.7 UPS Batteries 243
5.7.1 Vented (Flooded or Wet Cell) UPS Batteries 243
5.7.2 Valve Regulated (VRLA) UPS Batteries 244
5.7.3 Modular Battery Cartridge (MBC) UPS Batteries 245
5.7.4 Comparison of Three Common UPS Battery Technologies 245
5.7.5 Battery Monitoring 245
5.8 Summary 247
5.9 Review Questions 247
6 Power Distribution Unit and Cabling 250
6.1 Introduction 250
6.1.1 Basic PDU 250
6.1.2 Metered PDU 251
6.1.3 Switched PDU 251
6.2 Rack Power Distribution Unit and Redundancy 251
6.3 Power Feed to 42RU Rack 254
6.4 Data Center Power Cabling Installation 255
6.4.1 Transformation of the Data Center 255
6.4.2 Under the Floor Cabling 256
6.4.3 Overhead Cabling 257
6.5 Power Cable Layout Architectures 257
6.5.1 Star Topology Cabling Architecture 257
6.5.2 Bus Topology Cabling 258
6.6 Data Center Power Calculation 258
6.6.1 Process of Calculating Data Center Power Requirements 260
6.7 Strategies for Power Saving 265
6.7.1 Improve Efficiency of UPS or Remove Redundant Power Equipment 265
6.7.2 Improve Power Configuration 265
6.7.3 Reducing Data Center Capacity 267
6.8 Summary 269
6.9 Review Questions 269
7 Data Center Cooling 272
7.1 Introduction 272
7.2 Understanding Cooling, Comfort, and Precision Cooling 272
7.2.1 Understanding Cooling 272
7.2.2 Comfort Cooling 273
7.2.3 Precision Cooling 273
7.2.4 Issues with Not Using Precision Cooling 274
7.2.5 Heat Sources in a Data Center 274
7.3 Temperature, Pressure, and Volume 275
7.3.1 Heat 275
7.3.2 Temperature 276
7.3.2.1 Dry-Bulb Temperature (DBT) 276
7.3.2.2 Wet-Bulb Temperature (WBT) 277
7.3.2.3 Dew-Point Temperature (DPT) 277
7.3.3 Humidity 277
7.3.3.1 Relative humidity 278
7.3.3.2 Absolute humidity 278
7.3.3.3 Humidity ratio 278
7.3.4 Relationship between Temperature and Humidity 278
7.3.5 The Psychometric Chart (Humidity Chart) 280
7.3.6 Refrigeration 281
7.3.7 Refrigeration Unit 282
7.3.8 Refrigeration Cycle 282
7.3.8.1 Evaporation (state 1) 283
7.3.8.2 Compression (state 2) 283
7.3.8.3 Condensation (state 3) 285
7.3.8.4 Expansion (state 4) 285
7.3.9 Airflow and Airfow Rate 285
7.3.9.1 Gas laws 286
7.3.9.2 Boyle’s law 286
7.3.9.3 Charles’ law 287
7.3.9.4 Gay-Lussac’s law 287
7.3.10 Fan Types and Fan Laws 287
7.3.10.1 Axial and propeller fans 288
7.3.10.2 Centrifugal and radial fans 289
7.3.10.3 Fan laws 289
7.4 Data Center Cooling Components 290
7.4.1 CRAC 290
7.4.2 CRAH 290
7.4.3 Chiller 290
7.4.4 Humidifier and Dehumidifier 290
7.5 Data Center Cooling Control 291
7.5.1 Demand Fighting among Different CRAC Units 292
7.5.2 Adopting a Dew Point and Avoiding Relative Humidity Control 293
7.5.3 How to Control Humidity and Temperature 294
7.5.4 Consequences of Under- or Overhumidification 294
7.5.5 Managing the Data Center Temperature 295
7.5.5.1 Rack temperature measurement 296
7.5.5.2 CRAC temperature measurement 296
7.5.5.3 ASHRAE thermal guidelines for controlling temperature 298
7.5.6 Making Temperature Changes Based on the Heat Transfer Equation 299
7.5.7 Five Different Technologies for Removal of Data Center Heat 300
7.5.7.1 Air cooled DX system (two piece) 300
7.5.7.2 Air-cooled self-contained system (one piece) 301
7.5.7.3 Ceiling mounted system 302
7.5.7.4 Glycol-cooled system 302
7.5.7.5 Water-cooled system 302
7.5.7.6 Chilled water system 303
7.6 Summary 304
7.7 Review Questions 307
8 Effective Air Distribution in Data Centers 308
8.1 Introduction 308
8.2 Methods of Air Distribution 309
8.2.1 Flooded Approach for Hard Floor 309
8.2.2 Targeted or Locally Ducted Approach for Hard Floor 310
8.2.3 Fully Ducted or Contained Approach for Hard Floor 310
8.2.4 Locally Ducted for Supply Air with Hard Floor 311
8.2.5 Fully Ducted for Both Supply and Return Air with Hard Floor 312
8.2.6 Locally Ducted or Targeted Approach with Raised Floor 313
8.2.7 Fully Ducted Return Air with Raised Floor 314
8.2.8 Fully Ducted Supply Air with Raised Floor 314
8.2.9 Fully Ducted Supply Air and Locally Ducted Return Air with Raised Floor 314
8.2.10 Fully Ducted Supply and Return Air with Raised Floor 315
8.3 Guidelines for Air Distribution Methods 316
8.4 Computational Fluid Dynamics (CFD) Analysis 317
8.4.1 What Is Data Center CFD Analysis and Simulation? 318
8.4.2 The Process of CFD Modeling and Simulation 319
8.5 Data Center Cooling Calculations 321
8.5.1 Converting Energy in kW to Tons of Ice Cooling Equivalent 321
8.5.2 IT Load Calculations 321
8.5.2.1 Assumptions 321
8.5.2.2 Cooling load calculations 322
8.5.3 Total Cooling Requirement Calculation 322
8.5.3.1 UPS heat output calculation 322
8.5.3.2 PDU heat output calculation 323
8.5.3.3 Light heat output calculation 323
8.5.3.4 People heat output 323
8.5.3.5 Summary of all heat outputs 324
8.5.3.6 Other consideration for cooling requirements 324
8.5.3.7 High density blade server cooling considerations 324
8.6 Managing and Optimizing Cooling Systems 326
8.6.1 Resolve Easy Issues Immediately to Improve Cooling Efficiency 326
8.6.1.1 Install blank panel 326
8.6.1.2 Manage racks and cables properly 326
8.6.1.3 Optimizing raised floor height for cooling 327
8.6.2 Guidelines to Manage Perforated Tiles and Racks 328
8.6.2.1 Avoid the Venturi effect 330
8.6.2.2 Avoid the supply short circuit 331
8.6.3 Conditional Monitoring for Cooling System 332
8.6.4 Handling High-Density Rack Cooling 332
8.6.4.1 Row-based and rack-based cooling 333
8.6.4.2 Cold and hot aisle containment 334
8.6.4.3 Summary of pros and cons of different containment approaches 335
8.6.4.4 Which one is better? 337
8.7 Summary 338
8.8 Review Questions 338
9 Cooling Strategy 340
9.1 Cooling Control for Wiring Closets 340
9.1.1 Sharing Comfort Cooling System 340
9.1.2 Conduction Cooling 341
9.1.3 Conduction, Passive, and Fan-Assisted Ventilation 342
9.2 Room-Based Cooling 342
9.3 Row-Based Cooling 343
9.4 Rack-Based Cooling 344
9.5 Comparison of Room-, Row-, and Rack-Based Cooling 344
9.5.1 Mixing with Room and Row Based Cooling 345
9.5.2 Hot Aisle and Rack Containment for High-Density Zone 347
9.5.3 Uncontained 348
9.6 Rack Rear Door–Based Cooling Strategy 348
9.7 Raising the Data Center Temperature 349
9.8 Free Cooling Using Economizers 351
9.8.1 Airside Economizer 353
9.8.2 Waterside Economizer 353
9.9 Summary 357
9.10 Review Questions 362
10 Fire Suppression and On-Site Security 364
10.1 Introduction 364
10.2 Issues with Traditional Fire Suppression Systems 365
10.3 Fire Classification and Standards 366
10.3.1 Fire Detection 366
10.3.1.1 Computer room detection 367
10.3.1.2 Power room detection 367
10.3.1.3 Fire detection system 368
10.4 Fire Suppression Solution Selection 368
10.4.1 Traditional Fire Suppression Solutions 370
10.4.1.1 Carbon dioxide (CO2) fire suppression 370
10.4.1.2 Water-based (or water mist) fire suppression 370
10.4.1.3 Halon 372
10.5 Inert Gases, Halocarbons, and Aerosol 373
10.5.1 Inert Gases 373
10.5.2 Halocarbons 373
10.5.3 Aerosol 373
10.5.4 Fluorinated Ketone (Liquid) (Novec 1230) 374
10.5.5 Most Commonly Used Agents in Today’s Data Center 374
10.6 Fire Suppression System Cost for Data Centers 374
10.7 Summary of Fire Suppression Selection 375
10.8 On-Site or Physical Security 377
10.9 Physical Layers 379
10.9.1 Protecting Data Center Perimeters 379
10.9.2 Security Envelope 381
10.9.3 Access Points and Door Control 381
10.9.4 Camera or CCTV Control 382
10.9.5 Security Guards 382
10.10 Organizational Layer 383
10.10.1 People 383
10.10.2 Organizational Structure and Policy 385
10.10.3 Security Process 386
10.11 Establishing Physical Security 386
10.11.1 Cost Calculations for Physical Security Systems 387
10.11.1.1 Proportion of data center infrastructure cost 388
10.11.1.2 Cost per watt per month (opex)+capex 388
10.11.1.3 Cost per terabytes data storage (opex)+capex 389
10.11.1.4 Baseline cost plus incremental opex per square meter of computer room 389
10.11.2 Summary of physical security 389
10.12 Summary 389
10.13 Review Questions 390
III. Cloud Infrastructure and Management 392
11 Cloud Infrastructure Servers: CISC, RISC, Rack-Mounted, and Blade Servers 394
11.1 Cloud Servers 394
11.1.1 A Client/Server Architecture 398
11.2 x86 Server 400
11.2.1 CPU 404
11.2.1.1 Socket 404
11.2.1.2 Chip 405
11.2.1.3 Core, multicore, processor, and CPU 405
11.2.1.4 N-way servers 407
11.2.1.5 Multithreading and processes 408
11.2.1.6 Hyperthreading 409
11.2.2 Server CPU Cache 410
11.2.3 RAM 410
11.2.4 NUMA 411
11.2.5 Server PCI Cards 413
11.2.6 Server Storage 414
11.2.7 Server Network 415
11.2.8 Server Motherboard 415
11.3 Rack-Mounted Servers and Vendors 415
11.4 Blade Servers 418
11.4.1 What Is a Blade Server? 418
11.4.2 History of Blade Servers 420
11.4.3 Rack vs. Blade Server 424
11.5 RISC Server 425
11.5.1 History of RISC Servers 426
11.5.2 CISC vs. RISC 427
11.5.3 RISC Server Market Share 431
11.6 Oracle/Sun SPARC Servers 432
11.6.1 Oracle/Sun M-Series RISC Servers 437
11.6.2 Oracle/Sun T-Series RISC Servers 440
11.6.3 SPARC Logical Domain and Virtual Machine (VM) 441
11.7 Summary 446
11.8 Review Questions 447
12 Cloud Storage Basics 448
12.1 Storage Hierarchy 448
12.1.1 Hard Disk Drive (HDD) Fundamentals 449
12.1.1.1 HDD physical metrics 450
12.1.1.2 HDD evolution 453
12.1.2 Storage SLA and RAID Architecture 456
12.1.2.1 The common definition of an SLA 457
12.1.2.2 RAID techniques 458
12.1.2.2.1 Striping 459
12.1.2.2.2 Mirroring 460
12.1.2.2.3 Parity 460
12.1.2.3 RAID configurations 460
12.1.2.3.1 RAID-0 460
12.1.2.3.2 RAID-1 461
12.1.2.3.3 RAID-5 (distributed parity with N+1) 461
12.1.2.3.4 RAID-6 (distributed parity with double parity redundancy) 462
12.1.2.3.5 RAID-10 or RAID-01 (nested RAID-1 and RAID-0 or RAID1+0) 462
12.1.2.4 Comparison of RAID options 463
12.1.2.4.1 Summary of common RAID characteristics, cost and write penalties 464
12.1.2.4.2 RAID options and application IOPS 465
12.1.3 Storage LUN 466
12.1.3.1 LUN capacity expansion 467
12.1.3.1.1 Meta LUN concatenation 467
12.1.3.1.2 Meta LUN striping 468
12.1.3.2 LUN masking 468
12.2 Solid State Disk or Flash SSD 468
12.2.1 What Is an SSD? 471
12.2.2 SSD versus HDD 473
12.2.3 Total Cost of Ownership of SSD 474
12.3 Storage Topologies and Connections 476
12.3.1 Direct Attached Storage (DAS) 476
12.3.1.1 Internal DAS 476
12.3.1.2 External DAS 476
12.3.2 Storage Area Network (SAN) 477
12.3.3 Network Attached Storage (NAS) and File Storage Protocols 482
12.3.3.1 The idea of NAS 482
12.3.3.2 Elements of a NAS device 483
12.3.3.2.1 Special server and network elements 484
12.3.3.2.2 Storage elements 484
12.3.3.2.3 Software elements and file system 485
12.3.3.2.4 Integrated NAS 485
12.3.3.2.5 Gateway NAS 487
12.4 Storage Protocols 487
12.4.1 File-Oriented Protocols 487
12.4.1.1 Server Message Blocks (SMB)/Common Internet File System (CIFS) 488
12.4.1.2 Network File System (NFS) 488
12.4.2 Block-Oriented Protocols 490
12.4.2.1 IDE/ATA/parallel ATA or PATA 490
12.4.2.2 Serial ATA or SATA 491
12.4.2.3 SCSI 493
12.4.2.4 ISCSI 495
12.4.2.5 Fibre Channel Protocol (FCP) 497
12.4.2.6 Fibre Channel IP (FCIP) 499
12.4.2.7 Internet Fibre Channel Protocol (iFCP) 500
12.4.2.8 Fibre Channel over Ethernet (FCoE) Protocol 500
12.4.2.8.1 Converged Network Adapter (CNA) 503
12.4.2.8.2 Fibre Channel over Ethernet (FCoE) Switch 503
12.4.3 Storage Interface Protocols Summary 503
12.5 Pros and Cons for Different Storage Topologies 506
12.6 Traditional Storage vs. Cloud Storage 509
12.7 Major Storage Vendors and Market Trends 513
12.8 Summary 516
12.9 Review Questions 517
13 Data Center Networks 520
13.1 Key Network Terms and Components 520
13.1.1 Network Hardware 521
13.1.1.1 Hub 522
13.1.1.2 Switch 525
13.1.1.3 Bridge 527
13.1.1.4 Router 532
13.1.1.4.1 Principles of routing 532
13.1.1.4.2 Router size 534
13.1.1.4.3 Router types 534
13.1.1.4.4 Routing protocols 536
13.1.1.5 Gateway 538
13.2 Data Center Network Terms and Jargon 539
13.2.1 DCN Terms, Jargon, and Definitions 539
13.2.1.1 Topology 540
13.2.1.2 Network topology 540
13.2.1.3 Data center network topology 540
13.2.1.4 Node 541
13.2.1.5 Node degree 541
13.2.1.6 Neighbor nodes 541
13.2.1.7 Diameter 541
13.2.1.8 Dimension 541
13.2.1.9 Radix 541
13.2.1.10 Regular topology 542
13.2.1.11 Irregular topology 542
13.2.1.12 Nonblocking and blocking 542
13.2.1.13 Direct network 542
13.2.1.14 Indirect network 542
13.3 Metrics of DCN Topology 543
13.4 Types of Network Topology 544
13.4.1 Common DCN Topologies 550
13.4.1.1 Basic trees 551
13.4.1.2 Fat tree 552
13.4.1.3 Commodity switch fabric-based fat tree (Al-Fares) 553
13.4.1.4 Top of Rack (ToR) solution 555
13.4.1.5 End of Row (EoR) and middle of ROW (MoR) solutions 556
13.4.2 Recursive DCN Topologies 558
13.4.2.1 DCell 558
13.4.2.1.1 Principles of DCell 558
13.4.2.1.2 Structure of DCell 559
13.4.2.1.3 DCell formula 561
13.4.2.1.4 DCell summary 561
13.4.2.2 BCube 562
13.4.2.2.1 Principles of BCube 562
13.4.2.2.2 Structure of BCube 563
13.4.2.2.3 BCube formula 564
13.4.2.2.4 BCube summary 566
13.4.3 Other DCN Topologies 567
13.4.3.1 Virtual layer 2 (VL2) 567
13.4.3.1.1 Principles of VL2 568
13.4.3.1.2 Structure of VL2 570
13.4.3.1.3 Summary of VL2 571
13.4.3.2 Conventional butterfly and flattened butterfly 572
13.4.3.2.1 Principle of flattened butterfly 573
13.4.3.2.2 Structure of flattened butterfly 574
13.4.3.2.3 Flattened butterfly formula 575
13.4.3.2.4 Flattened butterfly summary 576
13.4.3.2.5 2-Dilated Flattened Butterfly (2DFB) 578
13.4.3.3 Dragonfly topology 578
13.4.3.3.1 Principle of dragonfly solution 580
13.4.3.3.2 Structure of dragonfly 580
13.4.3.3.3 Dragonfly formula 581
13.4.3.3.4 Dragonfly summary 582
13.4.4 Characteristics of Different DCN Topologies 583
13.5 Characteristics of Cloud Data Center Network 583
13.5.1 Management Network 583
13.5.2 Kernel Network 584
13.5.3 Virtual Machine Network 588
13.5.4 Virtualized Storage Network 588
13.5.5 Example of Connection Details 589
13.6 Cloud DCN Summary 593
13.6.1 DCN Component Summary 593
13.6.2 Terms and Definitions Summary 596
13.6.3 Metrics Summary 596
13.6.4 DCN Topology Summary 596
13.6.5 Cloud DCN 598
13.7 Review Questions 599
IV. Cloud Computing Cost Models and Framework 600
14 Cost Modeling: Terms and Definitions 602
14.1 Concept of Cost Model 603
14.1.1 Definition of Cost 603
14.1.1.1 Tangible costs 604
14.1.1.2 Intangible costs 604
14.1.1.3 Cost parameters 604
14.1.1.4 Sunk cost 606
14.1.1.5 Direct Variable Cost (DVC) 607
14.1.1.6 Capital Expenditure (Capex) 608
14.1.1.7 Operational Cost or Operational Expenditure (Opex) 608
14.1.2 Capex and Opex Shift in a Cloud Environment 608
14.1.3 Benefits 612
14.1.4 Risks and Opportunity 612
14.1.5 Definition of Model 613
14.1.5.1 Objective cost model 616
14.1.5.2 Subjective cost model 617
14.1.6 Model Measurement or Metrics 618
14.1.7 Analysis 620
14.1.8 Framework and Methodology 621
14.1.9 Formulating a Cost Model 622
14.2 Purposes of Cost Modeling for Cloud Computing 623
14.2.1 Visualize Abstract Structure of the Complex World 623
14.2.2 Organize Concepts, Thoughts, and Ideas 624
14.2.3 Communicate with Other People 625
14.3 Challenges of Cloud Cost Modeling 625
14.3.1 Not All Factors Are within the Framework 629
14.3.2 Limitation of Framework Size 630
14.3.3 Objective or Subjective Process of Cost Modeling 630
14.3.4 Limitation of Individual Knowledge and Experience 630
14.3.5 A Time Stamp on the Model 631
14.4 Summary 631
14.5 Review Questions 632
15 Cost Model Categories 634
15.1 Review of Cost Models 634
15.1.1 The Cost Model of the First CPU 638
15.1.2 Recent Cloud Computing Cost Models 639
15.1.2.1 Hybrid solution for cloud computing cost model 641
15.1.2.2 Cloud service provider’s cost model 645
15.1.2.3 Optimizing cost models 646
15.1.2.4 Cost model using the method of traditional economic mapping 648
15.1.2.5 Cost model oriented by service level agreement (SLA) 648
15.1.2.6 Cost model from a TCO perspective 652
15.1.2.7 Computable general equilibrium (CEG) model 655
15.2 Cloud Computing Issues, Impacts, the Right Questions for the Cost Model 655
15.2.1 Cloud Service Consumers 656
15.2.2 Cloud Service Providers 656
15.3 Cost Models over the Last 50 Years 656
15.4 Common Financial Cost Models 659
15.4.1 Accounting Rate of Return (ARR) 660
15.4.2 Breakeven Point (BEP) 661
15.4.3 Cost/Benefit and Cost/Benefit Ratio 662
15.4.4 Cost of Revenue Model 663
15.4.5 Internal Rate of Return (IRR) 663
15.4.5.1 What are the pros and cons of IRR? 664
15.4.6 Net Present Value (NPV) 664
15.4.6.1 What are the pros and cons of the NPV model? 665
15.4.7 Simple Payback Period (SPP) 665
15.4.8 Discounted Payback Period 666
15.4.9 Profitability index 666
15.4.10 Return on Investment (ROI) Model 667
15.4.11 Total Cost of Ownership 668
15.4.12 TCO/ROI Model 668
15.5 Summary 669
15.6 Review Questions 670
16 Chargeback 672
16.1 Introduction to Chargebacks 672
16.1.1 Understanding Enterprise IT Operations 674
16.2 No IT Cost Allocation 679
16.3 Non-IT-Based Cost Allocation 681
16.4 IT Domain–Based Cost Allocation 681
16.4.1 Direct Cost 682
16.4.2 Measured Resource Usage 682
16.4.3 Subscription-Based Cost Allocation 684
16.4.4 High-Level Allocation 684
16.4.5 Low-Level Allocation 685
16.4.6 Hardware-Based Cost 685
16.4.7 Static Capacity–Based Cost 685
16.4.8 Ticket-Based Cost 685
16.4.9 Peak Level–Based Cost 686
16.4.10 Virtual Server– or VM Account–Based Cost 686
16.5 Fee-Based Cost Allocation 687
16.5.1 Negotiated Flat Rate 687
16.5.2 Tiered Flat Rate 688
16.5.3 Transaction Ratio–Based Cost 688
16.5.4 Activity-Based Cost 689
16.5.5 SLA Performance Metrics 690
16.6 Business-Based Cost Allocation 691
16.6.1 Fixed Revenue-Based Cost 691
16.6.2 Fixed Revenue with Predefined Range 692
16.6.3 Profit-Oriented Cost Model 692
16.6.3.1 Capacity reservation–based rate 693
16.6.3.2 Bidding instance (market base rate) 693
16.7 Summary 693
16.8 Review Questions 694
V. Cloud Strategy and Critical Decision Making 696
17 Cost Model Calculation 698
17.1 Case Study 698
17.1.1 Company History 698
17.1.2 Basic Business Profile 699
17.1.3 Current IT Assets and Operation 702
17.1.4 Strategic IT Investment Decision Options 702
17.1.4.1 Data center facility capex 703
17.1.4.2 IT hardware expenses 704
17.1.4.3 Software licenses 704
17.1.4.4 Other implementation costs 704
17.1.4.5 Building an E2E cost framework 704
17.1.5 Model Assumption Details 708
17.1.5.1 Server workload assumptions 708
17.1.5.2 Server cost assumptions and vendor selection decision 708
17.1.5.3 Network cost assumptions 709
17.1.5.4 Storage cost assumptions 709
17.1.5.5 Data center facility cost assumptions 711
17.1.5.6 VMware hypervisor license cost assumptions 712
17.1.5.7 Operation system and other middleware assumptions 715
17.1.5.8 Amazon EC2 and S3 cost assumptions 715
17.2 Calculation Steps and Results 716
17.2.1 Calculate Growth Rate 718
17.2.2 Calculate Dedicated and Virtualized Workload 720
17.2.3 Calculate Static Net Present Value (NPV) 722
17.3 Conclusion of Case Study 724
17.4 Summary 726
17.5 Review Questions 728
18 Real Option Theory and Monte Carlo Simulation 730
18.1 Overview of Real Option Theory 730
18.2 History of Real Options 731
18.3 What Are Real Options? 733
18.3.1 Equations of Real Option Theory 735
18.3.2 Criteria of Real Options from a Project Perspective 735
18.3.3 Real Options for Investment Decision 737
18.3.3.1 Learning option 737
18.3.3.2 Modular or discrete option 737
18.3.3.3 Insurance option 738
18.3.3.4 Irreversible option 738
18.3.3.5 Flexible option 738
18.3.3.6 Platform option 738
18.4 Possible Real Options 739
18.4.1 Growth or Expansion or Leveraging Option 739
18.4.2 Time-to-Build or Open Option 740
18.4.3 Multiple Interacting Options 740
18.4.4 Option to Switch 740
18.4.5 Option to Defer 741
18.4.6 Option to Alter the Operating Scale 741
18.4.7 Option to Abandon (Put Option) 741
18.4.8 Different Terms for Real Options 741
18.5 Real Options versus Financial Options 742
18.6 Real Options versus Traditional Approaches 743
18.7 What Is Monte Carlo Simulation (MCS)? 747
18.7.1 Probability of Probability (Monte Carlo Tests) 749
18.7.2 Simulation Process 750
18.7.3 Different Types of Monte Carlo Simulation 753
18.7.3.1 Linear MCS 753
18.7.3.2 Nonlinear MCS 754
18.7.4 Pros and Cons of Monte Carlo Simulation (MCS) 755
18.7.4.1 What is MCS good at? (Pros) 755
18.7.4.2 What is MCS not good at? (Cons) 755
18.7.4.3 Good applications for MCS 756
18.7.4.4 Bad applications for MCS 757
18.8 Random Numbers and Brownian Motion 758
18.8.1 Pseudorandom versus Random Numbers 758
18.8.2 Brownian Motion (BM) and Geometric BM 759
18.8.2.1 Brownian motion 759
18.8.2.2 Wiener process or standard brownian motion 761
18.8.2.2.1 Levy Processes 761
18.8.2.2.2 Mathematical Terms of Standard Brownian Motion 762
18.8.2.2.3 Brownian Motion with Drift 763
18.8.2.2.4 Geometric Brownian motion 764
18.9 MCS and ROT Process 764
18.9.1 Calculation Process 764
18.9.2 Tactical Level of Analysis 766
18.9.2.1 Project prioritization and listing target projects for analysis 766
18.9.2.2 Static NPV calculation (Traditional Analysis) 767
18.9.2.3 Verifying business criteria 767
18.9.2.4 Monte carlo simulation for revenue forecasting 767
18.9.2.5 Checking that everything makes sense and MCS input calibration 768
18.9.3 Strategic Level of Analysis 769
18.9.3.1 Strategic level of real options problem 769
18.9.3.2 Real option modeling 769
18.9.3.3 Portfolio and resource optimization 770
18.9.3.4 Documenting conclusions and recommendations 771
18.9.3.5 Update and revise 772
18.10 Summary of MCS and ROT Concepts 772
18.11 MCS Analysis Process Details 774
18.11.1 Sensitivity Analysis with DCF (Five-Year Revenue) 774
18.11.1.1 Change Discount Cash Flow (DCF) ±10% 775
18.11.1.2 Change initial capex ±10% 775
18.11.1.3 Change interest rate ±10% 775
18.11.2 Sensitivity Analysis with Different Scenarios 775
18.11.3 Monte Carlo Simulation (MCS) Analysis 777
18.11.3.1 Scenario MCS (Radical Sensitivity) 777
18.11.3.2 Normal case MCS (General Sensitivity) 777
18.12 Real Option Theory Analysis Process Details 781
18.13 Real Option Theory Process Equations 782
18.13.1 Implement the Real Option Value Calculation Process 783
18.13.1.1 Step 1: Binomial lattice process 785
18.13.1.2 Step 2: forward process 785
18.13.1.3 Step 3: backward induction process 785
18.14 Summary 791
18.15 Review Questions 793
Appendices 796
Appendix A. Catalogue of Major IT Project Catastrophe 796
Appendix B. An Example of BRD Template 805
B.1 Business Requirements Document Table of Contents 806
Appendix C. Global Data Center Map (100 Countries and 3236 Data Centers for Colocation in 2014 Based on datacentermap.com) 808
Appendix D. Comparison of Different Cost Models 809
Appendix E. Nineteen Free Cloud Storage Options (2013 Data) 810
Appendix F. List of Different Cost Model Analysis 811
Appendix G. Server Products Provided by Major Server Different Vendors 812
G.1 IBM Rack-Mounted Server 812
G.2 Dell Rack-Mounted Server 813
G.3 Lenovo Rack-Mounted Server 813
G.4 Huawei Rack-Mounted Server 813
G.5 Oracle/Sun ×86 Rack-Mounted Server 814
G.6 Fujitsu Rack-Mounted Server 814
G.7 Cisco Rack-Mounted Server 815
Appendix H. TIA-942 Telecommunication Infrastructure Standard for Data Center Tier 815
References 818
Index 832

Textauszug

Preface

Caesar Wu and Rajkumar Buyya, Melbourne, Australia, 2014

How can we measure the sky? This question sometimes refers to how to measure the cost of cloud computing. For many people, it is a very challenging and tough question. And yet, many C-class senior executives (CEO, CFO, and CIO), stakeholders, and cloud investors would not only want to know “how” (cost model assumptions and calculations), but also want to know “why” (logic behind these assumptions).

Why is this so important? The simple answer is it is too big to be ignored. We have heard many stories about how some decision makers just throw big money into cloud projects without proper understanding of cloud technology and expect to catch up to the “wind” (win). This book will lay out the basic concepts and foundation of cloud computing and data center facilities and then provide tools and practical approaches for decision makers to make the right strategic investment decisions. It will help the decision maker to not only rely on “gut feelings” or previous experiences but also count on the scientific method.

One of the goals of this book is to establish a practical framework to enable IT executives to make a rational choice when they are facing a multimillion-dollar investment decision for a cloud project, which is to determine whether IT workloads should stay local or fly to a cloud. (inhouse or cloud computing).

Almost five years ago, this challenging task was assigned to us because a senior IT executive wanted to justify a multimillion investment decision that he had already made but he was not sure whether the decision was a rational choice or not. The original idea of this exercise was to check his intuition, estimate the strategic value, communicate with all the stakeholders, and change the scope of the cloud investment project if necessary.

At that time, many trial projects of cloud computing, server virtualization, and software multitenancy had just taken off. Various companies made different investment decisions in order to test the water or get a foothold on the cloud market.

With these intentions in our mind plus many years’ practical experience in cost modeling of utilities and grid computing, hosting services management, network design, construction, operation, lifecycles, and service delivery, we elicited eight basic questions about this cost modeling exercise:

• What is the ultimate goal of measuring the sky?

• How many cost models are there?

• How can we make a logical and rational comparison with different models?

• Why is the TCO/ROI model is so popular? If we use TCO/ROI, would it be the right choice?

• What are the assumptions of these models?

• How can I select the right model to fit a particular business need?

• How can we establish both revenue- and nonrevenue-based cost models?

• What are the risks of keeping the IT workload in house versus migrating to the cloud?

We believe that most people, whether they are cloud service providers or cloud service consumers, will also face similar questions if they are asked to measure “the sky” or to prepare a business case for a cloud investment project. From this perspective, this book is also targeted for IT business analysts and MBA students as reference material.

In essence, the core objective of this book is to demonstrate how to build a cloud cost model. It will illustrate the process of establishing the cost framework and calculating the costs. One of the main reasons to address the cloud cost modeling issue is that many ordinary people have two popular misconceptions:

1. The cloud is free.

2. My data is stored anyway up in the air.

If this is so, why should we bother to measure the sky? The answer is dependent on who you are. If you are just an individual consumer and require very limited cloud resources, it is quite clear that you can obtain nearly free cloud resources. However, if you are a business consumer, especially for medium- and large-scale businesses, there will be no free lunch. You have to pay for what you have consumed. This leads to the issue of how to make the rational investment decision for the usage of IT resources.

For most small or medium size companies, the investment decision would be relatively simple. The decision criteria could be mainly based on financial or economic returns plus a decision maker’s intuition or personal satisfaction. However, for a large enterprise, the strategic investment decision (very often involving millions of dollars) is not a simple intellectual exercise but rather than process of negotiation and compromise among different Line of Business (LoB) units.

However, to some degree, all models are subjective because cost modeling involves many subjective assumptions and selection of raw data and material. It would be impossible to avoid subjective assumptions and personal opinions. Strictly speaking, any data selected and assumption made are subjective. It is based on personal experiences and intuition or perhaps, a gut feeling.

Many people think a gut feeling is negative or nonscientific. As a matter of fact, a gut feeling is kind of a super-logic or sixth sense or recognition of a subconscious pattern. It gives us a shortcut to quickly reach a solution. Sometimes, this shortcut serves us quite well, especially if we do not have enough time to analyze the circumstances surrounding us or do not have enough information available. In this case, the sixth sense would be the only choice for us to reach a self-satisfactory conclusion. It is not purely arbitrary or an illogical guess but rather meta–knowledge built upon the subconscious mind. Actually, people’s minds are always searching for a recognised pattern based on available information, knowledge, experiences and most importantly, wisdom. Perhaps that is why a gut feeling is very often called an “educated guess,” self-learning, working experience, or armchair thinking.

Many strategic investment decisions made by IT legends such as Steve Jobs and Marc R. Benioff [1] led to great success for their companies. Why did they achieve what most people cannot achieve? Is it because they not only have years of working experiences and cumulative knowledge, but also have “gut feelings” or wisdom? People speculate that they may have absorbed wisdom from Eastern philosophy and religion because they both went to India for enlightenment. In Steve Jobs’ own words, “Trust in destiny” and “Follow your heart.” Walter Isaacson, the exclusive biographer of Steve Jobs, wrote it this way:

Jobs’s interest in Eastern spirituality, Hinduism (Krishna/God Consciousness), Zen Buddhism, and the search for enlightenment was not merely the passing phase of a nineteen-year-old. Throughout his life he would seek to follow many of the basic precepts of Eastern religions, such as the emphasis on experiential prajñā, wisdom or cognitive understanding that is intuitively experienced through concentration of the mind. Years later, sitting in his Palo Alto garden, he reflected on the lasting influence of his trip to India [2].

For the East, it is the soul. The soul did not come with body nor die with the body. The body is just a temporary home for the soul. The soul can be enlightened by many sophisticated methodologies and practices that have been developed by Eastern philosophy, religion, and culture for many thousands of years or by messages delivered by the Supreme God personally (e.g., Lord Krishna’s teachings compiled as Bhagavad Gita) or his incarnations.

For the West, it is subconsciousness. In Sigmund Freud’s teachings, it is the unconscious mind beneath consciousness and awareness. It is a repository of idea, desire, memories, and emotion. It consists of any information and data the mind collects from five senses but cannot consciously process to make meaningful sense of. However, it can be retrieved or recalled to consciousness by the simple direction of attention.

In order to make the right decision at the right time, the spiritual mind constantly needs not only information and knowledge but also wisdom. Without that, a strategic decision may just be a tactical one. Long-term success would be dependent on pure luck rather than a strategy. Here, wisdom means abstract pattern recognition at hierarchical level. It is the experience of cumulative knowledge. Cumulative knowledge has four different levels:

• Level 1: You do not know what you do not know (ignorance).

• Level 2: You know what you do not know (know unknowns).

• Level 3: You know what you know and what you do not know (know your boundaries).

• Level 4: You know all – knowledge of knowledge or meta-knowledge, wisdom (wizard).

For many people and under many circumstances, they are just wandering around atknowledge level 1. If we borrow the Indian philosophy term, it is so-called “ignorance.” There are two different scenarios when people face the unknown. One is either leaving to chance or pretending to know. The other is to wonder about the unknown and continuously search for knowledge and wisdom. That is why people often say wondering is the beginning of wisdom.

Unfortunately, we have witnessed many IT strategic decisions made by some wayward people subject to purely static...

Produkt-Details

• Erscheint lt. Verlag: 27.2.2015
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik ► Netzwerke
• Themenwelt: Mathematik / Informatik ► Informatik ► Web / Internet
• ISBN-10: 0-12-801688-4 / 0128016884
• ISBN-13: 978-0-12-801688-6 / 9780128016886