Combined architecture of the configuration management system for computing infrastructure

Abstract

Information systems and services of any computing infrastructure require configuration in accordance with the current tasks assigned to them. Currently, configuration management systems are used to configure infrastructure nodes, which make it possible to automate this process. One of the basic principles of these systems is the presence of a control node in the infrastructure, from which configuration management will be performed, on which the configuration files themselves will also be located. The main difference lies in the principle of interaction with managed nodes. In one case, the management node is responsible for initializing the node configuration, sending configuration files to managed machines, which will then be configured according to this configuration, this management model is called the push model. In another case, the pull model is used, which provides that the managed node is responsible for initializing the request. It independently polls the control node for changes in its configuration. Usually, all infrastructure nodes have different network access policies: some are constantly available and their status is known to the system administrator, while others are under the full control of their users and it is impossible to say with certainty when they will appear on the network and receive the configuration. In accordance with this, a system architecture is proposed that implements both management models, since in this case it will not be necessary to use several configuration management systems to configure all nodes in the infrastructure.