Survey of software configuration management tools of nodes in heterogeneous distributed computing environment

Abstract

The paper discusses the problem of preparing a computing environment for large-scale scientific experiments in the process of continuous integration of applied and system software. A comparative analysis of software configuration management tools (such as Chef, Ansible, Puppet, and SaltStack) of computational nodes in a heterogeneous environment is being performed. These tools are intended for automating the configuration of different nodes. Such automation reduces the setup time of nodes and increases the reliability of computations by minimizing the number of software and hardware failures, associated with the human factor in the manual configuration process. For the development of scientific applications, the Orlando Tools framework is used. Based on the results of the comparative analysis and requirements of this framework, the Ansible framework was selected for further integration into the chain of continuous integration of applied and system software. Practical experiments have shown the advantages of using Ansible in comparison with other systems of a similar purpose.