作者: Christer Carlsson , Robert Fullér
DOI: 10.1007/978-3-642-15976-3_2
关键词: Probabilistic logic 、 Node (networking) 、 Grid computing 、 Risk assessment 、 Statistical model 、 Grid 、 Task (computing) 、 Data mining 、 Granularity 、 Computer science
摘要: We developed a hybrid probabilistic and possibilistic technique for assessing the risk of an SLA computing task in cluster/grid environment. The probability success with model is estimated higher than since takes into consideration possibility distribution maximal number failures derived from resource provider’s observations. showed that we can increase or decrease granularity as needed; reduce estimate P(S*=1) by making rougher, more conservative, unlikely events (M+1, N) node failures. noted M which dependent on history nodes being used be calibrated to “a few” “many” nodes.