ABSTARCT

Fault Tolerance in CORBA

Area of Specialization : Distributed Systems

Year : 1998

Place : Computer Science Department, “Politehnica” University of Bucharest

Master of Science Thesis

The thesis is structured in six chapters and presents the necessary notions to understand the fault-tolerance concepts.

The first chapter is an introduction in the current status of distributed environments and their problems, today. It emphasizes the importance of a fault-tolerant system for deploying high-available applications.

The second chapter explains the object replication process. The possible implementations for a fault-tolerant system are shortly described. The three approaches: integration approach, interception approach and service approach are presented withexpressive figures, supported by explanations.

The third chapter lists the features necessary for a system to be fault-tolerant and provide reliable communication. The thesis suggests several solutions for some of the characteristics.

The main part is the fourth chapter, that suggests some mechanisms for fault-tolerance. The object replication based methods are already implemented by Electra system. It provides services for object replication and multicast communication. Each model is supported by pseudo-code with the main instructions in the programs of the client, the system and the server.

Several implementation of CORBA compliant systems are presented in the fifth chapter. Most of major software companies have created their own version of CORBA compliant system. Each system is shortly described. More information can be found on Internet, at the URL address from bibliography.

The last chapter reminds the importance of fault-tolerant systems for high-available applications.