Scholarlink

This study presents the development of an intelligent system for predicting node failures in Kubernetes clusters using machine learning techniques, specifically neural networks. The purpose of the model is to provide maintaining cluster stability and reliability, which is especially important in conditions of high service availability requirements. The model is trained on data including node operational parameters, system metrics, and event logs. It allows identifying potential failures at early stages. The paper proposes a machine learning algorithm that includes several stages of data processing: collecting resource metrics and event logs, pre-processing them to extract significant features, and training to predict node failures in a Kubernetes system. This solution helps DevOps engineers automate the monitoring process and take preventive measures to eliminate problems, which significantly reduces downtime and increases overall system performance. The implementation of this model represents a significant step forward in the field of Kubernetes-based infrastructure management, improving service processes and reducing the risk of container-related emergencies.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Failure Prediction in Kubernetes Systems Using Neural Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Eldor S. Elmurodov,
  • Denis N. Medvedev,
  • Yury E. Karyakin,
  • Ivan Yu. Karyakin

摘要

This study presents the development of an intelligent system for predicting node failures in Kubernetes clusters using machine learning techniques, specifically neural networks. The purpose of the model is to provide maintaining cluster stability and reliability, which is especially important in conditions of high service availability requirements. The model is trained on data including node operational parameters, system metrics, and event logs. It allows identifying potential failures at early stages. The paper proposes a machine learning algorithm that includes several stages of data processing: collecting resource metrics and event logs, pre-processing them to extract significant features, and training to predict node failures in a Kubernetes system. This solution helps DevOps engineers automate the monitoring process and take preventive measures to eliminate problems, which significantly reduces downtime and increases overall system performance. The implementation of this model represents a significant step forward in the field of Kubernetes-based infrastructure management, improving service processes and reducing the risk of container-related emergencies.