Ms. Renukadevi Chuppala
Using Univariate Linear Regression to Analyze and Manage Repository Storage Space
A repository manager is an essential tool in DevOps and CI/CD pipelines, commonly used for managing, storing, and retrieving binary artifacts efficiently. One such tool is designed to handle artifacts, dependencies, and containers, playing a critical role in software development by centralizing storage and streamlining management. This system facilitates collaboration and ensures that teams can maintain control over their software supply chain. On a daily basis, various artifacts are uploaded to multiple repositories within each instance of the system, and the space consumed increases proportionally with the volume of artifacts. It is the administrator's responsibility to clear the space when it exceeds a certain threshold. If the space is not cleared, it can cause the server to shut down, affecting the entire operation. Administrators can use automated tasks to assist with cleanup, but these tasks have limitations, particularly when it comes to deleting only unnecessary artifacts. If valuable artifacts take up too much space, it becomes impossible to predict when the server might crash. This paper offers a solution to this problem by utilizing machine learning techniques, specifically Linear Regression Analysis, to predict space usage. By creating a regression model based on historical data, it allows for the prediction of future space usage, enabling administrators to better manage and allocate resources before reaching critical levels.