Safe Stopping of Running Component-Based Distributed Systems: Challenges and Research Gaps

Abstract

Continuous availability of services and low degree of disruption are two inherent necessities for mission-critical software systems. These systems could not be stopped to perform updates because disruption in their services consequent irretrievable losses. Additionally, compared to offline update, the changes should preserve the correct completion of ongoing activities. In order to place the affected elements in a safe state before dynamic changes take place, the notion of tranquility has been proposed to make quiescence criterion less disruptive and easier to obtain. Additionally, some other approaches have been proposed in order to tackle the shortcomings of these seminal proposals. However, these approaches impose some challenges to the safe dynamic reconfiguration of component-based systems. In this paper, existing challenges to preserve global consistency during runtime software reconfiguration in distributed contexts are described. The contribution of this paper is to propose a number of guidelines which can be served as agenda for future direction of research to enable a dependable safe stopping of running component-based systems.