摘要:In this work, we present an integral scheduling systemfor non-dedicated clusters, termed CISNE-P, which en-sures the performance required by the local applications,while simultaneously allocating cluster resources to pa-rallel jobs. Our approach solves the problem efficientlyby using a social contract technique. This kind of tech-nique is based on reserving computational resources,preserving a predetermined response time to local users.CISNE-P is a middleware which includes both a pre-viously developed space-sharing job scheduler and adynamic coscheduling system, a time sharing schedul-ing component. The experimentation performed in aLinux cluster shows that these two scheduler compo-nents are complementary and a good coordination im-proves global performance significantly. We also com-pare two different CISNE-P implementations: one de-veloped inside the kernel, and the other entirely imple-mented in the user space
关键词:space and time sharing scheduling;coscheduling; social contract