Grids, P2P and Services Computing

Preface

Contents

List of Contributors

XtreemOS: a Sound Foundation for Cloud Infrastructure and Federations

1 XtreemOS

Scalability.

Transparency.

Towards a Grid File System Based on a Large-Scale BLOB Management Service

1 Introduction

2 The building blocks: Gfarm and BlobSeer

2.1 The Gfarm grid file system

2.1.1 Overview of Gfarm’s architecture

2.2 The BlobSeer BLOB management service

2.2.1 BlobSeer at a glance

2.2.2 Overview of BlobSeer’s architecture

2.3 Why combine Gfarm and BlobSeer?

3 Towards an object-based file system based on Gfarm and BlobSeer

3.1 How to couple Gfarm and BlobSeer?

3.2 The Gfarm/BlobSeer file system design

4 Experimental evaluation

5 Conclusion

References

Improving the Dependability of Grids via Short-Term Failure Predictions

1 Introduction

2 Analyzing Grid Tomography Data

2.1 The Anatomy of a Grid

2.2 The FailBase repository

2.3 Types of monitoring data

2.4 Preprocessing

2.5 Modeling methodology

3 Experimental Results

3.1 Evaluation metrics: recall and precision

3.2 Analysis of prediction accuracy

3.2.1 Selecting the target attributes

3.2.2 Data characteristics and accuracy

3.2.3 Effects of different classification algorithms

4 Conclusions

Acknowledgements

References

Distributed Data Mining using a Public Resource Computing Framework

1 Introduction

2 A Distributed Framework for Public Computing with Caching Capabilities

2.1 Parallel Mining of Closed Frequent Itemset

3 Performance Evaluation

4 Conclusions

References

Integration of the ProActive Suite and the semantic-oriented monitoring tool SemMon

1 Introduction

2 Related work

3 Extending SemMon’s support for specific platforms

4 ProActive integration strategy

4.1 Features of the systems to be integrated

4.2 ProActive requirements for monitoring vs. SemMon

4.3 Concept of architecture

4.4 Extensions introduced into the integrated system

5 Sample monitoring session

5.1 SemMon’s impact on monitored system

6 Conclusions and Future work

References

An Experimental Evaluation of the DQ-DHT Algorithm in a Grid Information Service

1 Introduction

2 Background on DQ-DHT

2.1 Broadcast over a Chord DHT

2.2 The DQ-DHT algorithm

3 Experimental evaluation

3.1 System prototype

3.2 Experimental results

4 Related work

5 Conclusions

Acknowledgement

References

Reducing traffic in DHT-based discovery protocols for dynamic resources

1 Introduction

2 Related Work

2.1 MAAN Range Query Implementation

3 REMED

3.1 Reducing the Routing Overhead

3.2 Skipping Publication

4 Cost model

4.1 The Standard MAAN Model

4.2 The REMED Optimized Model

5 Simulations

5.1 Tools and Implementation

5.2 Data Analysis

5.3 Cache Evaluation

5.4 Churn Evaluation

5.5 Model Evaluation

6 Conclusion and Future Work

References

Autonomic management of multiple non-functional concerns in behavioural skeletons

1 Introduction

2 Parallel framework

3 Autonomic management of multiple concerns in structured parallel computations

3.1 Centralized vs. distributed autonomic management of multiple concerns

3.2 Shared knowledge among different autonomic managers

3.3 Impact of local decisions on global application management

3.3.1 Consensus building

3.3.2 Consensus results

3.4 Initialization of theAM hierarchy

3.5 Rule-based multi-concern autonomic manager implementation

4 Sample case study

5 Related work

6 Conclusions

References

Decision Models for Resource Aggregation in Peer-to-Peer Architectures

1 Introduction

2 Related Work

3 The resource aggregation setup

4 Decision Models for Resource Aggregation

4.1 Onicescu

4.2 Promethee

4.3 The Multi-Attribute Utility Theory

5 Experiments and results

6 Conclusion

References

MappingWorkflows on Grid Resources: Experiments with the MontageWorkflow

1 Introduction

2 Background

3 A Motivating Example

4 Experimental Evaluation

4.1 The Simulator and Settings

4.2 Results and Discussion

4.2.1 All tasks at each level have a similar (estimated) execution time

4.2.2 All tasks at each level, except the rightmost task, have an (estimated) execution time which is twice the (estimated) exec

4.2.3 The leftmost task at each level has an (estimated) execution time, which is 10 times the (estimated) execution time of the

4.2.4 Summary

5 Conclusion

References

A Proposal on Enhancing XACML with Continuous Usage Control Features*

1 Introduction

2 Background

2.1 UCON

2.2 XACML

3 U-XACML policy syntax and semantics

4 U-XACML architecture

5 A case study

6 Conclusions

References

Self-* and Adaptive Mechanisms for Large Scale Distributed Systems

1 Introduction

2 Elastic management of computational Grid resources

3 Bio-inspired algorithms for self-organizing distributed systems

4 Non-intrusive monitoring and healing of complex applications

5 Self-management in desktop Grids

6 Conclusions

Acknowledgements

Network Monitoring in the age of the Cloud

1 Introduction

2 LAN virtualization

2.1 The IEEE802.1Q protocol

3 The VLAN-cloud connection

3.1 Amazon EC2

3.2 VMware

4 Network Monitoring in the age of the Cloud

4.1 Network monitoring on the user side

4.2 Network monitoring on the service provider side

5 Conclusions

5.1 Looking ahead

References

Acronyms

Recent Advances and Research Challenges in Desktop Grid and Volunteer Computing

1 Introduction

2 History and Classification of Desktop Grid Systems

3 Towards Data Desktop Grid

4 Integrating Desktop Grids into Cyber-Science Infrastructures

5 Conclusion

References

Research Challenges in Managing and Using Service Level Agreements

1 Introduction

2 SLA Research Landscape

3 Electronic Markets

4 Smart Licencing

5 WS-Agreement

5.1 Initialisation of the negotiation process

5.2 Negotiation of the template

5.3 Post-processing of the templates

6 Future Directions and Research Challenges

6.1 Conclusions and Research Directions

References