Transforming cloud-native application monitoring with neural network for anomaly detection

New generation cloud applications have significantly transformed software landscapes with scalable, elastic and robust solutions. However, as the number of microservices and distributed systems increase the monitoring and detection of these anomalies becomes rather difficult. In traditional MMSs, the workloads are not dynamic and thus does not capture any real-time problems hence a delay in responding to critical problems. The contribution of this paper is a new solution to improve cloud-native application monitoring by using neural networks for the same purpose. To ascertain presumptive exceptions, our method taps on deep learning models to process multivariate telemetry data in real-time. There is also an intention to accommodate high dimensional and noisy data as are characteristic of cloud-native applications to afford better detection accuracy and fewer false positives as embodied in the proposed framework. We support this proposition with a detailed experimental evaluation on real-world datasets for the purpose of illustrating its practical applications in improving reliability and utilization of resources and reducing down time. This paper lays down a roadmap toward better, wiser, and more anticipative monitoring solutions for cloud-native environments, thus opening the way for more dependable and self- healing systems.