Towards Highly Available and Self-Healing Grid Services

Stefania Costache (1) , Thomas Ropars (1) , Christine Morin (1)

1 MYRIADS - Design and Implementation of Autonomous Distributed Systems

The volatility of nodes in large scale distributed systems endangers the availability of grid services and makes them difficult to use. In such a context, structured peer-to-peer overlays can be used to provide scalable and fault tolerant communication mechanisms. To ensure the availability of services, active replication can be used on top of the overlays. In this paper, we present Semias, a framework that is based on active replication on top of a structured overlay to provide high availability and self-healing for stateful grid services. The self-healing mechanisms of Semias ensure the high availability of the replicated services while minimizing the number of reconfigurations. We have used Semias to make Vigne grid middleware services highly available. Experiments run on Grid'5000 and PlanetLab show the performance and self-healing properties of the framework.

Mots clés: Large scale distributed system; Grid Computing; High Availability; Self Healing

Domaines: Calcul parallèle, distribué et partagé [cs.DC]

Stefania Costache, Thomas Ropars, Christine Morin. Towards Highly Available and Self-Healing Grid Services. [Research Report] RR-7264, INRIA. 2010. ⟨inria-00476276⟩