|
||||||||||||||||||
|
||||||||||||||||||
|
One often neglected measure of robustness is the timeliness of system operation. For some systems, a late result is worse than no result at all. Not only does the late result contribute no value, but its computation tied up resources which could have been used by another computation to maximize the benefit to the system. Such systems are called hard real-time systems [1].
At the other extreme are the non-real-time systems in which timeliness is not a concern. It is sufficient that the computation returns a result. In between are soft real-time systems in which a late result has varying levels of utility. All computations can be classified somewhere on the continuum from non-real-time to hard real-time.
While we are not currently studying real-time systems, we receive frequent requests for these papers and hence make them available here as a service to the computing community.
[1] A real-time system is not necessarily a fast system. The defining characteristic is that the system strives to be timely. It only needs to be "fast enough" for the task at hand. Thus, a system which ensures that each frame of a movie is displayed within 30 milliseconds of the last one is just as much a real-time system as a process control system which requires 5 microsecond latencies.
Operated by the
University of California for the National Nuclear Security Administration, of the US Department of Energy. Copyright © 2001 UC | Disclaimer/Privacy |
|
NOTICE: Information from this server resides on a computer system funded by the U.S. Department of Energy. Anyone using this system consents to monitoring of this use by system or security personnel. For complete conditions of use see Disclaimer/Privacy. |