Available:*
Library | Item Barcode | Call Number | Material Type | Item Category 1 | Status |
---|---|---|---|---|---|
Searching... | 30000010193867 | QA76.9.E94 P476 2008 | Open Access Book | Book | Searching... |
On Order
Summary
Summary
With the fast development of networking and software technologies, information processing infrastructure and applications have been growing at an impressive rate in both size and complexity, to such a degree that the design and development of high performance and scalable data processing systems and networks have become an ever-challenging issue. As a result, the use of performance modeling and m- surementtechniquesas a critical step in designand developmenthas becomea c- mon practice. Research and developmenton methodologyand tools of performance modeling and performance engineering have gained further importance in order to improve the performance and scalability of these systems. Since the seminal work of A. K. Erlang almost a century ago on the mod- ing of telephone traf c, performance modeling and measurement have grown into a discipline and have been evolving both in their methodologies and in the areas in which they are applied. It is noteworthy that various mathematical techniques were brought into this eld, including in particular probability theory, stochastic processes, statistics, complex analysis, stochastic calculus, stochastic comparison, optimization, control theory, machine learning and information theory. The app- cation areas extended from telephone networks to Internet and Web applications, from computer systems to computer software, from manufacturing systems to s- ply chain, from call centers to workforce management.
Table of Contents
Preface | p. v |
About the Editors | p. ix |
List of Contributors | p. xv |
Part I Performance Design and Engineering | |
1 Machine Learning Techniques-Reductions Between Prediction Quality Metrics | p. 3 |
1.1 Introduction | p. 3 |
1.2 Basic Definitions | p. 4 |
1.3 Importance-Weighted Classification | p. 5 |
1.4 Multiclass Classification | p. 10 |
1.4.1 One-Against-All | p. 10 |
1.4.2 Error-Correcting Coding (ECOC) Approaches | p. 13 |
1.4.3 Approaches Based on Pairwise Comparisons | p. 17 |
1.5 Cost-Sensitive Classification | p. 17 |
1.6 Predicting Conditional Quantiles | p. 20 |
1.7 Ranking | p. 25 |
1.8 Conclusion | p. 26 |
References | p. 26 |
2 Performance Engineering and Management Method - A Holistic Approach to Performance Engineering | p. 29 |
2.1 Background | p. 29 |
2.2 What is Performance Engineering? | p. 31 |
2.3 Overview of PEMM | p. 32 |
2.4 PEMM Theme - Requirements and Early Design | p. 33 |
2.4.1 Requirements and Performance | p. 34 |
2.4.2 Early Design and Performance | p. 35 |
2.5 PEMM Theme - Volumetrics | p. 36 |
2.5.1 Business Volumes | p. 36 |
2.5.2 Technical Volumes | p. 37 |
2.6 PEMM Theme - Estimation and Modeling | p. 38 |
2.6.1 Performance Estimating Techniques | p. 39 |
2.6.2 Selection of Performance Estimating Methods | p. 41 |
2.7 PEMM Theme - Technology Research | p. 42 |
2.8 PEMM Theme - Design, Development and Tracking | p. 42 |
2.8.1 Recognizing Performance Patterns and Anti-Patterns | p. 43 |
2.8.2 Designing for Performance | p. 44 |
2.8.3 Performance Budgeting | p. 44 |
2.8.4 Performance Debugging and Profiling | p. 45 |
2.8.5 Design, Development and Tracking Guidance | p. 46 |
2.9 PEMM Theme - Test Planning and Execution | p. 46 |
2.10 PEMM Theme - Live Monitoring and Capacity Planning | p. 47 |
2.10.1 Relating PEMM to Performance and Capacity Management | p. 49 |
2.10.2 PEMM and ITIL Capacity Management | p. 50 |
2.11 PEMM Theme - Performance and Risk Management | p. 51 |
2.11.1 Assignment of Dedicated Performance Engineering Resources | p. 52 |
2.11.2 Applying PEMM to IT Project Governance | p. 52 |
2.11.3 Applying PEMM to Complex Projects | p. 53 |
2.12 Summary | p. 54 |
References | p. 55 |
3 Economic Models of Communication Networks | p. 57 |
3.1 Introduction | p. 57 |
3.1.1 General Issues | p. 58 |
3.1.2 Paris Metro Pricing | p. 60 |
3.2 Pricing of Services | p. 63 |
3.2.1 Tragedy of the Commons | p. 64 |
3.2.2 Congestion Pricing | p. 66 |
3.2.3 When to Use the Network? | p. 66 |
3.2.4 Service Differentiation | p. 69 |
3.2.5 Auctions | p. 73 |
3.3 Investment Incentives | p. 78 |
3.3.1 Free Riding | p. 78 |
3.3.2 Network Neutrality | p. 81 |
3.3.3 Economics of Security | p. 84 |
3.4 Conclusions | p. 86 |
References | p. 87 |
4 Algorithmic Methods for Sponsored Search Advertising | p. 91 |
4.1 Introduction | p. 91 |
4.2 Existing Auctions | p. 93 |
4.2.1 Practical Aspects | p. 95 |
4.3 The Advertiser's Point of View: Budget Optimization | p. 97 |
4.3.1 Modeling a Keyword Auction | p. 99 |
4.3.2 Uniform Bidding Strategies | p. 104 |
4.3.3 Experimental Results | p. 105 |
4.3.4 Extensions | p. 105 |
4.4 The Search Engine's Point of View: Offline Slot Scheduling | p. 106 |
4.4.1 Special Case: One Slot | p. 108 |
4.4.2 Multiple Slots | p. 110 |
4.4.3 Extensions | p. 113 |
4.5 The User's Point of View: a Markov Model for Clicks | p. 114 |
4.5.1 A Simple Markov User Click Model | p. 116 |
4.5.2 Properties of Optimal Assignments for Markovian Users .117 | |
4.5.3 Computing the Optimal Assignment | p. 118 |
4.6 Open Issues | p. 118 |
4.7 Concluding Remarks | p. 119 |
4.8 Acknowledgements | p. 120 |
References | p. 120 |
Part II Scheduling and Control | |
5 Advances in Oblivious Routing of Internet Traffic | p. 125 |
5.1 Introduction | p. 125 |
5.2 The Need for Traffic Oblivious Routing | p. 127 |
5.2.1 Difficulties in Measuring Traffic | p. 127 |
5.2.2 Difficulties in Dynamic Network Reconfiguration | p. 128 |
5.3 Traffic Variation and Performance Models | p. 128 |
5.3.1 Unconstrained Traffic Variation Model | p. 128 |
5.3.2 Hose Constrained Traffic Variation Model | p. 129 |
5.4 Oblivious Routing under Unconstrained Traffic Model | p. 130 |
5.5 Oblivious Routing of Hose Constrained Traffic | p. 131 |
5.6 Two-Phase (Oblivious) Routing of Hose Constrained Traffic | p. 132 |
5.6.1 Addressing Some Aspects of Two-Phase Routing | p. 135 |
5.6.2 Benefits of Two-Phase Routing | p. 137 |
5.6.3 Determining Split Ratios and Path Routing | p. 138 |
5.6.4 Protecting Against Network Failures | p. 139 |
5.6.5 Generalized Traffic Split Ratios | p. 140 |
5.6.6 Optimality Bound for Two-Phase Routing | p. 141 |
5.7 Summary | p. 143 |
References | p. 144 |
6 Network Scheduling and Message-passing | p. 147 |
6.1 Introduction | p. 147 |
6.2 Model | p. 150 |
6.2.1 Abstract formulation | p. 150 |
6.2.2 Scheduling algorithms | p. 151 |
6.2.3 Input-queued switch | p. 152 |
6.2.4 Wireless networks | p. 153 |
6.3 Characterization of optimal algorithm | p. 155 |
6.3.1 Throughput optimality | p. 156 |
6.3.2 Queue-size optimality | p. 159 |
6.4 Message-passing: throughput optimality | p. 165 |
6.4.1 Throughput optimality through randomization and message-passing | p. 165 |
6.4.2 Performance in terms of queue-size | p. 169 |
6.5 Message-passing: low queue-size or delay | p. 172 |
6.5.1 Input-queued switch: message-passing algorithm | p. 172 |
6.5.2 Wireless scheduling: message-passing scheduling | p. 178 |
6.6 Discussion and future direction | p. 182 |
References | p. 183 |
7 Introduction to Control Theory And Its Application to Computing Systems | p. 185 |
7.1 Introduction | p. 185 |
7.2 Control Theory Fundamentals | p. 186 |
7.3 Application to Self-Tuning Memory Management of A Database System | p. 191 |
7.4 Application to CPU Utilization Control in Distributed Real-Time Embedded Systems | p. 197 |
7.5 Application to Automated Workload Management in Virtualized Data Centers | p. 201 |
7.5.1 Introduction | p. 201 |
7.5.2 Problem statement | p. 203 |
7.5.3 Adaptive optimal controller design | p. 203 |
7.5.4 Experimental evaluation | p. 205 |
7.6 Application to Power and Performance in Data Centers | p. 207 |
7.6.1 Design Methodology for Integrating Adaptive Policies | p. 207 |
7.6.2 Evaluation | p. 211 |
7.7 Conclusions And Research Challenges | p. 212 |
References | p. 214 |
Index | p. 217 |