Architecture Based Approach to Adaptable Fault Tolerance in Distributed Object-Oriented Computing

Abstract

To gain high level of performance in distributed object oriented computing, a required level of reliability in objects has to be maintained. This brings in a set of complex requirements into consideration. Furthermore depending on the unpredictability of the underlying environment, the replication should have architecture for the adaptable fault tolerance so that it can handle different situations of the underlying system before the system fails. We propose a mechanism for analyzing the complexity of this underlying environments and designing a dynamically reconfigurable architecture. The architecture provides the user required reliability by analyzing the performance and the reliability of the underlying environment and then either adjusting the replication degree or adaptively shifting to a suitable replication protocol. This architecture is a part of the Juice system which supports adaptation properties for a distributed environment.