Fault classification in the architecture of virtual machine using deep learning

The performance of a network primarily depends on the probability of failure occurrence and its availability for various services, such as mitigation, latency gap, and simulations. Frequent faults in the cluster networks may result in task failure related to identifying and detecting these services. Therefore, detecting and classifying such faults and initiating corrective actions is required before they transform into system failure. We present a model that includes feature selection, an attention transformer, and feature transformer for fault classification. Our proposed model deals with tabular data with neural nets. The experimental analysis is carried out on the tabular dataset taken from the Telstra cluster network. The results have been reported, including failure records of service disruption events and total connectivity interruptions. The trace-driven experiments have been observed on the efficacy of our proposed model with an average of accuracy (98.3%), precision (98.3%), recall (97.4%), and F1 score (97.8%). The results validate the research objective to predict the failure occurrence in virtual machines.