ICNP97 Full Day Tutorial 2

High-performance Distributed Object Computing with CORBA

Douglas C. Schmidt
Washington University, St. Louis

Abstract

Developing robust, extensible, reusable, and efficient distributed systems and applications is hard. This tutorial presents an overview of distributed object computing based on CORBA 2.0. CORBA is an emerging open standard for distributed object computing that automates many common application development tasks such as service registration, location, and activation, demultiplexing, framing and error-handling, parameter (de)marshalling, and operation dispatching.

The tutorial describes how to significantly simplify and enhance the development of applications that use CORBA effectively in concurrent and distributed environments. This material is based on the presenter's extensive experience applying CORBA to production software projects at Motorola IRIDIUM, Kodak, Siemens, Bellcore, Ericsson, and McDonnell Douglas.

In addition, the tutorial presents techniques for measuring and optimizing the performance of CORBA Object Request Brokers (ORBs) and the CORBA Internet Inter-ORB Protocol (IIOP). This material is based on results from the presenter's research on high performance, real-time ORBs, which has appeared in top conferences such as SIGCOMM, ICDCS, INFOCOM, and GLOBECOM and are available at http://www.cs.wustl.edu/~schmidt/corba-research.html/.

Topics Covered

The topics covered in this full-day tutorial include the following:

• Overview of CORBA (2 hours)
  • Object Request Broker (ORB)
  • Internet Inter-ORB Protocol (IIOP)
  • Interface definition language (IDL)
  • Client stubs and skeletons
  • Object Adapters
  • Dynamic invocation
  • Object references
  • Object activation

• Advanced CORBA Features and services (1 hour)
  • Multi-threaded servers
  • CORBA Services (e.g., event service, real-time services, concurrency service)
  • Integration with other communication frameworks

• CORBA performance and optimizations (2 hours)
  • Measurements pinpointing CORBA bottlenecks over low-speed and high-speed networks
  • Optimization techniques for CORBA ORBs and IIOP, such as:
    • Streamlining for the common case
    • Eliminating gratuitous data copying presentation layer overhead
    • Replacing unwieldy general purpose functions by smaller, more efficient ones
    • Adding redundant protocol state to speed up expensive operations
    • Reducing concurrency overhead
  • Supporting real-time Quality of Service (QoS) in CORBA
    • Real-time ORB scheduling and dispatching mechanisms
    • Real-time CORBA event service

• Related work (1 hour)
  • Comparison of CORBA with related distributed object frameworks like DCOM and DCE

Speaker Biography

Douglas Schmidt is an Assistant Professor of Department of Computer Science and Radiology at Washington University in St. Louis, Missouri, USA. His research focuses on design patterns, implementation, and experimental analysis of object-oriented techniques that facilitate the development of high-performance, real-time distributed object computing systems on parallel processing platforms running over high-speed ATM networks. Dr. Schmidt has published widely in IEEE, ACM, IFIP, and USENIX technical conferences and journals on topics ranging from high-performance communication software systems and parallel processing for high-speed networking protocols to distributed object computing with CORBA and object-oriented design patterns.