As fault tolerance is the ability of a system to perform its function correctly even in the presence of faults. Therefore, different fault tolerance techniques are critical for improving the efficient utilization of expensive resources in high performance data grid systems. One of the most popular strategies of fault tolerance is the replication, it creates multiple copies of resources in the system and it has been proved to be an effective way to achieve data availability and system reliability. In this paper the authors propose a new adaptive dynamic replication that combines between a replication based on availability and replication based on popularity. The authors' adaptive dynamic replication uses two types of replicas (primary and ordinary) and two types of placement nodes (best client and best responsible nodes) for the new replicas. In addition to the replication, we used other strategies such as fault detection, fault prediction, dynamicity management, self-stabilization. All these services are grouped in one fault tolerance box named Collaborative Services for Fault Tolerance (CSFT) that structure them in hierarchical services and organize the relationships between them.
Performance Evolution of Dynamic Replication in a Data Grid using DMDR Algorithm