| Title |
Machine Learning based Failure Prediction Method for Supporting Optimal System Reliability |
| Authors |
서형준(Hyungjun Seo) ; 노재춘(Jaechun No) ; 박성순(Sungsoon Park) |
| DOI |
https://doi.org/10.5573/ieie.2021.58.1.41 |
| Keywords |
Failure prediction; Machine learning; Optimization; System monitoring; Docker |
| Abstract |
As IT services become more common, the dependence of computer systems in various companies has increased, and many studies have been conducted on predicting failures to ensure system availability. Existing studies have tried to predict system failures in advance through machine learning, and in this process, various techniques such as data analysis, preprocessing, and optimization are applied. However, due to the lack of data, existing disability prediction research data can use a limited number of machine learning models in the process of generating artificial data or generating prediction models, and specialized knowledge of machine learning is required. In addition, a large cost was incurred by analyzing and comparing performance through iterative learning. In this paper, to solve this problem, an optimal model is created by comparing the performance of various models in the optimization process, and machine learning models can be easily added, and a system failure prediction frame that automates the optimization process and the overall process of machine learning. I implemented the work. The framework collects system information through a monitoring system, and through it, performs optimization processes such as feature selection and hyper parameter tuning to generate a predictive model. The predictive model uses real-time data measured through the monitoring system to determine whether there is a system failure and to generate an alarm. |