Large Scale Network-Centric Distributed Systems

22

Virtualized Environment Issues in the Context of a Scientific Private Cloud

Bruno Schulze, Henrique de Medeiros Klôh, Matheus Bousquet Bandini, Antonio Roberto Mury, Daniel Massami Muniz Yokoyama, Victor Dias de Oliveira, Fábio André Machado Porto, and Giacomo Victor McEvoy Valenzano

Contents

22.1 Introduction

22.2 Related Works

22.3 Methodology

22.3.1 Experiments and Objectives

22.3.2 Experimental Infrastructure

22.4 Experiments

22.4.1 Experiment 1: Influence of the Hypervisors on Performance

22.4.2 Experiment 2: Hybrid Virtualized Environment Evaluation

22.4.3 Experiment 3: Virtualized Database

22.4.4 Result Analysis

22.5 Conclusion

22.6 Glossary

Acknowledgments

References

22.1 Introduction

Science and technology are becoming increasingly dependent on resources with high computing power and storage capacity. The growing development of research activities in various areas of science and industry are generating even larger and more complex data sets, making the use of high-performance computing essential to data analysis, storage, and the quick and efficient sharing of these data and results.

Cloud computing has emerged as a tool to provide access to resources meeting these demands. However, to accomplish this goal, there are still several difficulties related to information security, availability, and performance.

In this study, we focus on aspects related to performance in private Clouds, in support to scientific applications, verifying the influence of the hypervisor used, ...