Applying Patterns to Improve Performance of Fault-Tolerant CORBA

There is a significant trend to develop mission-critical, embedded, telecommunications, and financial distributed systems based on distributed object computing middleware, such as CORBA. Applications for these systems often require the underlying middleware, operating systems, and networks to provide end-to-end quality of srvice (QoS) support to enhance their efficiency, predictability, scalability, and fault tolerance. The Object Management Group (OMG), which standardizes CORBA, has addressed many of these application requirements recently in the Real-time CORBA and Fault-tolerant CORBA specifications. This paper provides two contributions to the study and design of CORBA middleware that provides multiple QoS properties. First, we describe results of experiments conducted to measure the performance of a fault-tolerant CORBA services framework called DOORS and illustrate how common implementation pitfalls can adversely affect performance. Second, we describe the patterns we are incorporating into the DOORS fault-tolerant CORBA service and the TAO CORBA ORB to simultaneously improve its performance and fault-tolerance.