Architecture and implementation method of real-time recovery in distributed systems
Joshua Etkin, John H. Choi · Microprocessing and Microprogramming · 1991
Continuous operation is an important factor in real-time environment. In the case of a client, server or even link failure, the system should recover as fast as possible and maintain its operation if permissible. The paper introduces a new recovery mechanism that operates in conjunction with the existing recovery manager. The proposed recovery solution, named Cart Wheel Looping Mechanism, is comprised of various processing protocols and interconnected architecture. Through the invocation of these protocols, various types of failures are detected and corrected, so the system can meet its real-time constraints. The paper advocates the incorporation of the new recovery mechanism on top of an existing recovery manager, so it will be activated only when the existing manager falls short of supporting the real-time requirements.