Proactive Monitoring of IT System’s Critical Parameter

Abstract

The new IT structures demand active monitoring systems that have the ability to identify abnormalities in important parameters of the system, before the failures affect the performance and the reliability. This paper introduces a new framework called the UACP-PM (Uncertainty-Aware Contrastive Prototype-based Proactive Monitoring) to monitor the IT system metrics intelligently through time-series learning and adaptive decision mechanisms. The suggested framework starts with preprocessing multi-source telemetry data by filtering noise, eliminating outliers, imputing missing values, and normalizing. A time-series building method based on a sliding window is used to learn time-based dependencies, after which a Temporal Convolutional Network with a dual attention mechanism is used to extract features. The TS2Vec (Time Series to Vector) contrastive representation learning strengthens the features, and a prototype-guided transformer classifier is used to identify fault states. Evidential deep learning is built in to measure uncertainty of prediction and enhance reliability of alerts. Lastly, real-time monitoring and constant adjustment is made possible by visualization dashboards. Experimental assessment shows that the suggested framework has an anomaly detection rate of 98.2% with a precision of 98.1, recall of 98.0 and an F1-score of 98.2, and an AUC value of 99.9, thus minimizing false alarms and enabling the sound proactive decision-making in dynamic IT settings.