Anomaly identification and failure diagnosis

Abstract

When we work in a large scale network, number of problems arises, the total time required to deal with these type of problems depends on how severe the problem is? As system takes more time to recover from failures, maintenance cost goes on increasing, it also causes loss of processing cycles. To deal with such type of loss, the information at various nodes in network is collected and verification of failure reasons is performed. In traditional system this process of dealing with failures was handled by humans, but such a manual processing was leading to various problems such consumption of time, scalability of network and many more. As scalability of network goes on increasing we should think on automation of anomaly identification to perform failure diagnosis.