REAL-TIME IN THE REAL WORLD
Lui Sha

Real time computing directly affects a nation's competitiveness and security. Given the globalization of the economy, it is more important than ever to quicken the pace in modernizing industrial computing facilities. To stay competitive, industry must be able to respond to changing needs much faster than can be achieved with traditional approaches. Examples of application areas in which real time computing problems arise include: * agile manufacturing facilities that are able to quickly reconfigure plant operations to meet changing requirements and to permit on-line maintenance and upgrade, * Multi-media applications across the networks, tele-conferencing, and distributed real time simulations. * sensor systems for monitoring the development of weather patterns, seismic data, the status of power distribution grids and the distribution of pollutant, * vehicular traffic control and air traffic control, * patient monitoring, heart-lung machines, CAT scanners, MRI, and other advanced medical equipment. * surveillance, command and control and other defense systems.

Compared with business and scientific computing, real time computing systems are often distributed and have distinctively stringent timing, reliability and safety requirements imposed by the application environment. For systems with very high availability requirements such as tele-communication network, early warning sensor network and power generation network, software and hardware maintenance must be done on-line without shutting the system down. These additional requirements makes real time computing systems a very challenging problem. To date, these requirements have typically been met by using proprietary hardware, software and then a long period of "fine tuning" by trial and error. Such a process creates systems that are expensive and difficult to maintain and upgrade. As a result, many real time systems used by industry and defense are often forced to keep their antiquated hardware and software, in spite of the rapidly fall cost of hardware.

From the prespective of R&D, part of the problem is also that real time problems are often treated as part of a specific application areas. For example, communication communities, OS communities and industrial control community often roll up specific solutions from a particular angle that often makes system integration difficult.

It is time to rise and meet the challenge of an increasingly competitive global economy. We need to create a forum where researchers from different application areas with real time concerns can come and together to create a coherent set of solutions. We must create a new generation of real time computing architecture, where computers, communication networks, process control and manufacturing technologies are organic parts of an intelligent, friendly, and responsive real time computing infrastructure. This new architecture must provide the framework in which real-time computing, fault tolerant computing, computer safety and security, high-performance networks and multi-media technologies can be effectively integrated with advanced process control and manufacturing technologies. The resulting new computing infrastructure should be a "living system" where new electro-mechanical equipments, new computers, new networks and new software can be easily introduced, leading to constant improvement in quality and efficiency in the production of goods and services.