List of Figures | p. ix |
List of Tables | p. xi |
Preface | p. xiii |
Contributing Authors | p. xix |
Trends in Parallel and Distributed Computing | p. 1 |
Trends in Processing Technology | p. 2 |
Dimension I - Advances in device technology | p. 3 |
Dimension II - Advances in concurrency technology | p. 4 |
Trends in Computer Networking Technology | p. 4 |
Network Component | p. 4 |
Transport Component | p. 6 |
Trends in Software Tools and Environments | p. 8 |
Parallel and Distributed Computing Environment | p. 13 |
Legion | p. 13 |
Legion Objectives | p. 13 |
Architecture | p. 14 |
Globus | p. 15 |
Globus Objectives | p. 15 |
Architecture | p. 16 |
Condor | p. 18 |
Condor Objectives | p. 20 |
Architecture | p. 20 |
Charlotte | p. 22 |
Charlotte Objectives | p. 22 |
Architecture | p. 22 |
Trends Toward Virtual Computing | p. 25 |
Application Computing Requirements | p. 25 |
Application Classes | p. 25 |
Characteristics of Representative Algorithms | p. 27 |
Parallelism Unit | p. 27 |
Computational Scalability | p. 27 |
Interconnection Requirement | p. 29 |
Control Requirement | p. 29 |
Natural Support Architecture | p. 29 |
Communication Protocol Requirements | p. 30 |
Simplicity | p. 30 |
Low latency and high bandwidth | p. 31 |
Concurrent Interprocess Communication | p. 31 |
Multicasting | p. 31 |
Support for Process Synchronization | p. 32 |
Efficient Flow Control Scheme | p. 32 |
Adaptive Distributed Virtual Computing Environment | p. 35 |
Overview of ADViCE Architecture | p. 35 |
Visualization and Editing Services (VES) | p. 37 |
Application Editing Service (AES) | p. 37 |
Application Visualization Service (AVS) | p. 39 |
ADViCE Server | p. 40 |
Initial and Runtime Mapping Service (IRS) | p. 41 |
Application Management Service (AMS) | p. 41 |
Application Control Service (ACS) | p. 43 |
Application Data Service (ADS) | p. 44 |
ADViCE Adaptation Approach | p. 45 |
Resource Allocation and Scheduling | p. 53 |
Overview of current scheduling algorithms | p. 53 |
Dynamic Level Scheduling | p. 55 |
Levelized Min Time | p. 55 |
Mapping Heuristic | p. 56 |
Heterogeneous Earliest Finish Time | p. 56 |
Definitions | p. 57 |
Definitions of Terms | p. 57 |
Definition of Task Scheduling | p. 59 |
Architecture of a Task Schedule | p. 59 |
Initial Mapping Service | p. 59 |
Runtime Mapping Service | p. 60 |
Analysis and Prediction of Task Execution | p. 61 |
Definitions and Notations | p. 61 |
Assumptions | p. 62 |
CPU Allocated Time--t[subscript CPU](T[subscript i], r[subscript j], 0, t) | p. 63 |
Task Execution Time Prediction--t[subscript rem](T[subscript i], r[subscript j], t) | p. 67 |
Future Load Prediction | p. 67 |
Tuning Function--t[subscript tun](T[subscript i], r[subscript j], t[subscript beg], t[subscript end]) | p. 70 |
Load Estimation Errors | p. 70 |
Tuning Function Example | p. 73 |
Task-Scheduling Algorithm | p. 79 |
Heterogeneous Earliest Finish Time (HEFT) Algorithm | p. 79 |
Task Prioritizing Phase | p. 79 |
Processor Selection Phase | p. 80 |
Complexity Analysis of the HEFT Algorithm | p. 80 |
HEFT Algorithm Drawbacks | p. 81 |
Dynamic Heterogeneous Prediction-Based Scheduling Algorithm (DHPB) | p. 82 |
The DHPB Algorithm for Initial Mapping Service | p. 83 |
The DHPB Algorithm for Runtime Mapping Service | p. 83 |
Complexity of DHPB Algorithm | p. 86 |
DHPB for Initial Mapping | p. 86 |
DHPB for Runtime Mapping | p. 86 |
Experimental Results and Analysis | p. 87 |
Experiment 1: Using ADViCE as a Parallel Evaluation Tool | p. 87 |
Experiment 2: Using ADViCE as a Problem Solving Environment | p. 92 |
Experiment 3: Evaluation of the ADViCE Adaptation Approach | p. 93 |
Experiment 4: Prediction of Task Execution Time | p. 96 |
Task-Scheduling Environment | p. 101 |
Experiment 5: Randomly Generated Application Flow Graph | p. 102 |
Experiment 6: Gaussian Elimination | p. 105 |
Summary and Future Directions | p. 109 |
Summary | p. 109 |
Future Work | p. 110 |
Future Trends | p. 111 |
References | p. 113 |
Index | p. 123 |
Table of Contents provided by Syndetics. All Rights Reserved. |