Predictability and Responsiveness of Fault Recovery Operations in Real-Time Systems (1997)
BibTeX
@MISC{Mejia-Alvarez97predictabilityand,
author = {Pedro Mejia-Alvarez and Juan A. De and La Puente},
title = {Predictability and Responsiveness of Fault Recovery Operations in Real-Time Systems},
year = {1997}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
This paper focuses on the problem of scheduling realtime and transient fault tolerant requests in a uniprocessor environment, using time redundancy. The problem of scheduling recovery time and replicate resources in a real-time system to enhance its fault tolerance, requires developing efficient scheduling policies, as well as fault tolerant mechanisms that resolve resource contention conflicts among different tasks requesting access to a shared resource such as the processor [2]. This is a complex problem because these scheduling policies and fault tolerant mechanisms must allow the scheduler to meet two competing objectives: ensure that the timing requirements of the real-time application workload are met, and ensure that the timing requirements of any recovery operation triggered by the detection of errors are met. While the scheduler strives to meet both of these objectives, there may be situations in which it may be unable to do so, these situations may be, (a) overload or high utilization of the real-time workload, (b) high rate of faults, and (c) unpredictability of fault recovery operations. An appropriate solutions must handle these conditions. Sources of unpredictability and bounds of execution must be identified and solved. Predictable mechanisms for fault tolerance must be developed, and schedulability analysis must be designed for integrating time-critical constraints of real-time workloads, and responsiveness-based priorities to recovery operations. All scheduling decisions must be performed considering that the timing correctness of the most critical real-time tasks is maintained and the state of the system is always predictable, even if some of the tasks are forced to suffer performance degradation. This paper describes ways to characterize the recovery ...
Keyphrases
real-time system fault recovery operation timing requirement fault tolerance recovery operation fault tolerant mechanism real-time workload resource contention responsiveness-based priority uniprocessor environment high rate recovery time time redundancy replicate resource real-time application workload transient fault tolerant request different task critical real-time task time-critical constraint timing correctness predictable mechanism appropriate solution performance degradation complex problem schedulability analysis high utilization scheduling decision
