DISTRIBUTED ARCHITECTURE FOR AN INTELLIGENT NETWORKING COPROCESSOR

摘要

Group communication technology, such as the Horus process, is used to implement a fault-tolerant high performance, reactive, real-time distributed IN coprocessor (14). The architecture of the distributed IN coprocessor (14) comprises workstation clusters, external adaptors (20a and 20b), and an update interface (24) interconnected by high speed communication links. Each workstation of the IN architecture represents a query element (22), so all the databases used by the IN coprocessor (14) in the course of servicing incoming request are split between query elements (22), provided that each of the workstation has access to the information stored in a certain database or databases. Group communication systems provide necessary features for managing and obtaining high reliability and operability of the IN coprocessor (14) including failure detection, notification of other members of the system about the failures, reconfiguration of the system to exclude failed members, bringing back into the system the members that have been recovered, and updating the recovered or new members according to the new state of the system.