Temporal Qos Management in Scientific Cloud Workflow Systems

by Liu; Chen; Yang

• ISBN13: 9780123970107
• ISBN10: 0123970105
• Format: Paperback
• Copyright: 2012-02-23
• Publisher: Elsevier Science

Summary

Cloud computing can provide virtually unlimited scalable high performance computing resources. Cloud workflows often underlie many large scale data/computation intensive e-science applications such as earthquake modelling, weather forecasting and astrophysics. During application modelling, these sophisticated processes are redesigned as cloud workflows, and at runtime, the models are executed by employing the supercomputing and data sharing ability of the underlying cloud computing infrastructures. Temporal QOS Management in Scientific Cloud Workflow Systems focuses on real world scientific applications which often must be completed by satisfying a set of temporal constraints such as milestones and deadlines. Meanwhile, activity duration, as a measurement of system performance, often needs to be monitored and controlled. This book demonstrates how to guarantee on-time completion of most, if not all, workflow applications. Offering a comprehensive framework to support the lifecycle of time-constrained workflow applications, this book will enhance the overall performance and usability of scientific cloud workflow systems. Explains how to reduce the cost to detect and handle temporal violations while delivering high quality of service (QoS) Offers new concepts, innovative strategies and algorithms to support large-scale sophisticated applications in the cloud Improves the overall performance and usability of cloud workflow systems

Table of Contents

Preface	p. ix
Acknowledgements	p. xi
About the Authors	p. xiii
Introduction	p. 1
Temporal QoS in Scientific Cloud Workflow Systems	p. 1
Motivating Example and Problem Analysis	p. 3
Motivating Example	p. 3
Problem Analysis	p. 5
Key Issues of This Research	p. 7
Overview of This Book	p. 8
Literature Review and Problem Analysis	p. 11
Workflow Temporal QoS	p. 11
Temporal Consistency Model	p. 12
Temporal Constraint Setting	p. 13
Temporal Consistency Monitoring	p. 14
emporal Violation Handling	p. 15
A Scientific Cloud Workflow System	p. 17
Novel Probabilistic Temporal Framework	p. 23
Framework Overview	p. 23
Component I: Temporal Constraint Setting	p. 26
Component II: Temporal Consistency Monitoring	p. 28
Component III: Temporal Violation Handling	p. 29
Forecasting Scientific Cloud Workflow Activity Duration Intervals	p. 33
Cloud Workflow Activity Durations	p. 33
Related Work and Problem Analysis	p. 35
Related Work	p. 35
Problem Analysis	p. 36
Statistical Time-Series-Pattern-Based Forecasting Strategy	p. 37
Statistical Time-Series Patterns	p. 38
Strategy Overview	p. 39
Novel Time-Series Segmentation Algorithm: K-MaxSDev	p. 41
Forecasting Algorithms	p. 43
Evaluation	p. 46
Example Forecasting Process	p. 46
Comparison Results	p. 50
Temporal Constraint Setting	p. 55
Related Work and Problem Analysis	p. 55
Related Work	p. 55
Problem Analysis	p. 57
Probability-Based Temporal Consistency Model	p. 58
Weighted Joint Normal Distribution for Workflow Activity Durations	p. 58
Probability-Based Temporal Consistency Model	p. 62
Setting Temporal Constraints	p. 64
Calculating Weighted Joint Distribution	p. 64
Setting Coarse-grained Temporal Constraints	p. 65
Setting Fine-grained Temporal Constraints	p. 66
Case Study	p. 68
Temporal Checkpoint Selection and Temporal Verification	p. 71
Related Work and Problem Analysis	p. 72
Related Work	p. 72
Problem Analysis	p. 72
Temporal Checkpoint Selection and Verification Strategy	p. 73
Probability Range for Statistically Recoverable Temporal Violations with Light-Weight Temporal Violation Handling Strategies	p. 73
Minimum Probability Time Redundancy	p. 74
Temporal Checkpoint Selection and Temporal Verification Process	p. 75
Evaluation	p. 76
Experimental Settings	p. 76
Experimental Results	p. 78
Temporal Violation Handling Point Selection	p. 81
Related Work and Problem Analysis	p. 81
Related Work	p. 81
Problem Analysis	p. 82
Adaptive Temporal Violation Handling Point Selection Strategy	p. 83
Probability of Self-Recovery	p. 83
Temporal Violation Handling Point Selection Strategy	p. 84
Evaluation	p. 85
Temporal Violation Handling	p. 95
Related Work and Problem Analysis	p. 95
Related Work	p. 95
Problem Analysis	p. 97
Overview of Temporal Violation Handling Strategies	p. 97
Temporal Violation Handling of Statistically Recoverable Temporal Violations	p. 98
Temporal Violation Handling of Statistically Non-Recoverable Temporal Violations	p. 99
A Novel General Two-Stage Local Workflow Rescheduling Strategy for Recoverable Temporal Violations	p. 100
Description of the General Strategy	p. 100
Metaheuristic Algorithm 1: GA	p. 103
Metaheuristic Algorithm 2: ACO	p. 105
Other Representative Metaheuristic Algorithms	p. 107
Three-Level Temporal Violation Handling Strategy	p. 107
PTDA for Level I Violations	p. 109
ACOWR for Level II Violations	p. 110
PTDA + ACOWR for Level III Violations	p. 110
Comparison of GA- and ACO-based Workflow Rescheduling Strategies	p. 112
Experimental Settings	p. 112
Experimental Results	p. 115
Evaluation of Three-Level Temporal Violation Handling Strategy	p. 122
Violation Rates of Local and Global Temporal Constraints	p. 122
Cost Analysis for a Three-Level Handling Strategy	p. 124
Conclusions and Contribution	p. 127
Overall Cost Analysis for Temporal Framework	p. 127
Summary of This Book	p. 128
Contributions of This Book	p. 130
Appendix: Notation Index	p. 133
Bibliography	p. 135
Table of Contents provided by Ingram. All Rights Reserved.