Detecting and tolerating failures in a loosely integrated heterogeneous database system

This paper presents the design and prototype implementation of a fault detection and fault tolerance facility based on the reactive system model for a loosely integrated heterogeneous database system running on a network of workstations. The database system consists of a number of fully autonomous databases that, through their willingness, can share some portion of their information. Sensors (functions that expose some part of the state of an object) are placed on individual databases and system objects (such as traders, agents, and other sensors). Other databases and system objects, and applications can subscribe to a sensor and therefore are notified whenever the sensor value changes. The fault detection mechanism uses sensors to monitor individual databases and system objects and to detect database or system faults. The fault tolerance mechanism uses actuators (functions that can change the state of an object) to react on database or system faults.