SCHEDULING SOFT REAL-TIME JOBS OVER DUAL NON-REAL-TIME SERVERS

In this paper, we consider soft real-time systems with redundant off-t he-shelf processing components (e.g., CPU, disk, network), and show ho w applications can exploit the redundancy to improve the system's abil ity of meeting response time goals (soft deadlines). We consider two s cheduling policies, one that evenly distributes load (Balance), and on e that partitions load according to job slackness (Chop). We evaluate the effectiveness of these policies through analysis and simulation. O ur results show that by intelligently distributing jobs by their slack ness amount the servers, Chop can significantly improve real-time perf ormance.