Institut numerique

Abstract

In the domain of cloud computing, especially the private cloud’s, current solutions for the IaaS (Infrastructure as a Service) implementation, which enable the administration and the use of virtual and physical resources within the network, using Web and virtualization technologies, remain based on centralized architectures for the monitoring of these resources. Moreover, these solutions don’t include fault-tolerance in their architectures. This may lead to the breaking of the services in case of failure of the system’s central node. In order to contribute to solve this problem, we have proposed a new system based on a distributed monitoring algorithm of the different machines on the network using the privilege turning principle. The solution uses a tree representation of the different nodes on the network to delegate child nodes of which the role is to oversee the critical parent nodes of the network. This way, an occurring failure at any point of these nodes is automatically detected by the elected child node which dynamically changes its state to get in charge of the functions and role of the leaving node. The system has been applied to IaaS services and has proved its ability to ensure the high availability of these services in a private cloud environment.

Key words: Virtualization, Private cloud computing, Distributed system, Distributed monitoring algorithm, Fault-tolerance.

Page suivante : Introduction générale

Retour au menu : CONCEPTION D’UNE SOLUTION DE CLOUD COMPUTING PRIVE BASEE SUR UN ALGORITHME DE SUPERVISION DISTRIBUE : APPLICATION AUX SERVICES IAAS