Graceful Degradation through Elasticity in Mixed-Criticality Federated Scheduling
Graceful Degradation through Elasticity in Mixed-Criticality Federated Scheduling
复制标题
混合关键联合调度中通过弹性实现优雅降级
DOI:
--
复制
发表时间:
2018
期刊:
影响因子:
--
通讯作者:
Steven Harris
中科院分区:
文献类型:
--
作者:
C. Gill;James Orr;Steven Harris
Mixed-criticality federated scheduling (MCFS) lets parallel real-time tasks with different criticality levels co-exist on a common multi-core host, guaranteeing that lower-criticality tasks do not interfere with timely completion of higher-criticality tasks, even if tasks exceed their nominal utilizations. However, lower-criticality tasks may be suspended or dropped in MCFS, to accommodate any higher-criticality task’s increased utilization requirements. Recent advances in elastic scheduling theory and concurrency platforms for parallel real-time (PRT) tasks offer a more graceful way to handle such resource reallocation, however, by exploiting elasticity of lower-criticality tasks’ utilizations. In this paper we summarize those advances, provide a combined elastic multi-criticality PRT task model, sketch strategies for leveraging elasticity, and give brief illustrative examples showing how such elasticity can provide a more graceful means of resource reallocation in mixed-criticality federated scheduling. Keywords—mixed-criticality, parallel real-time systems, elastic scheduling, federated scheduling