Support Fault-tolerant Corba Event Service Design And Implementation

During the past 10 years, Distributed Object Technology has developed very fast and has been widely applied in many domains such as manufacture, finance, telecommunication, insurance and traffic. CORBA is a specification of Distributed Object Technology. As it's independence of network protocols, programming languages and software platform, it has become more and more important in building large heterogeneous distributed system. CORBA specification provides basic object services that can be reused. These object services effectively reduce the cost of software development, configuration and management. CORBA event service is one of CORBA core services. CORBA event service provides decoupled, asynchronous communication mechanism, which builds a universal distributed communication model.With the ever increasing amount of CORBA applied in many critical domains, the requirement of high reliability and high availability of CORBA become more and more urgent. Fault-tolerant is an effective way to provide high reliability. Although OMG specified FT-CORBA specification in early 2001, and many institutes are doing research in FT-CORBA, CORBA products supporting fault-tolerant are still very few. Therefore, when developing reliable application based on CORBA middleware, the developer still need do a lot work on fault-tolerant largely in the application layer.This article describes the design and implementation of a fault-tolerant event service, referring to the author's research and development in a power monitoring system. Due to the requirement of power monitoring system, this event service using push-push model to transmit data. It provides functions such as event subscription, event subscription canceling and event filtering. The event service adopts an object redundancy manner to achieve fault-tolerant purpose. There are several replicas of event service and they inhabit different hosts. These replicas are peers one another, but when a client requests the event service, only one of the replicas executes the request. After a replica executes a request of the client, it must consist the other replicas with its change of state using CORBA method invocations. Besides, the replica with the latest startup time is responsible for maintaining all replicas consistency from time to time. Currently, the event service has been tested and put into use in power monitoring system.Since this event service relates close to power monitoring system, it is difficult to be applied to other system. By referring to the FT-CORBA model of ACE Tao, this...