Design principles and optimizations for high-performance, real-time CORBA

Applications and services for next-generation distributed systems must be reliable, flexible, reusable, and capable of providing low latency to delay-sensitive applications (such as avionics, and telecommunication systems), and high bandwidth to bandwidth-intensive applications (such as medical imaging, satellite surveillance, and teleconferencing) running over high-speed networks. Requirements for reliability, flexibility, and reusability motivate the use of object-oriented middleware like the Common Object Request Broker Architecture (CORBA).; However, the empirical studies we conducted measuring the performance of CORBA implementations revealed that current CORBA implementations incur a number of over-heads stemming from excessive presentation layer conversions, data copying, and inefficient server demultiplexing techniques.; This dissertation describes our work implementing a high performance, real-time ORB called "The ACE ORB" (TAO). This dissertation provides three contributions to building a high performance, real-time ORB. First, we will describe the optimizations we developed to improve the performance of the IIOP interpretive marshaling engine. Second, we will describe the efficient demultiplexing strategies we developed to reduce latency while improving scalability, predictability and consistency of request demultiplexing. Finally, we describe the design and implementation of the OMG IDL compiler we developed that generates efficient stubs and skeletons.