LAUSR

In complex networks, availability evaluation performs unsatisfactorily because the quality of service (QoS) is constrained by performance cost. This causes conventional evaluation only focuses on the failure probability of network devices as to indiscriminately maintain service at the same expense, leading to the performance cost increasing and available resource wasting. We developed a conceptually different evaluation approach, a Markov Chain with QoS based Network Availability (MCQNA), to identify the minimum performance cost of various businesses according to QoS metrics. Based on the analysis of limit probability of MCQNA, the values of the steady distribution were computed as the dynamic weight coefficients in performance cost. This paper details the MCQNA design and evaluates its performance in normal and heavy workloads. The evaluation results indicate that MCQNA with QoS indices judging performance cost is more reasonable and sensitive than the typical scheme. The study provided guidance for implementing strategy in networks.