Abstract
In this paper, we show how a distributed fault-tolerance component can be replaced dynamically. The need for such replacement arises due to the fact that there are often several fault-tolerance components that are suitable for adding faulttolerance to a fault, and the choice of the component depends upon the environment. Since a distributed fault-tolerance component has a fraction installed at every process, replacing such a component requires that the replacement be done consistently. More specifically, it requires that the dependency among the fractions of the component be handled so that a component fraction is not removed while other component fractions or the underlying application depends on it. We show how the dependency relation among component fractions is correctly handled while ensuring that the component replacement is transparent to the application. As an illustration of different types of dependency relations, we present a message communication example and show how dynamic composition is performed in it. Finally, in our approach, it is also possible to deal with faults that occur during dynamic composition.
Mahesh Arumugam
Mahesh Arumugam is a software engineer passionate about designing, programming, and deploying systems. Currently, I work in data security and analytics domain.