Research on the recovery methodology for special purpose information systems through the application of the infrastructure as code approach

This paper investigates the Infrastructure as Code (IaC) approach as a modern tool for automating the deployment, scaling and recovery of special-purpose information systems. The concept of separating the service lifecycle from the data lifecycle is substantiated as a key architectural prerequisite for IaC-oriented recovery: a failure at the ephemeral layer does not affect the integrity of the persistent layer, and system recovery reduces to recreating services from the repository followed by mounting the data storage. A synthetic comparative experiment was conducted for three recovery scenarios of a typical on-premise three-tier web information system: manual recovery using a Disaster Recovery Plan document, partial IaC, and full IaC approach. Each scenario was repeated five times; Student's t-criterion with a 95% confidence interval was applied to assess the reliability of the results. Based on the measurements, quantitative estimates were obtained for the Time to Restore Service, the Automation Coefficien, and the Recovery Point Objective. A mathematical formalization of IaC efficiency metrics is proposed, accounting for the statistical variance of measurements. It is established that the full IaC approach reduces the service recovery time to 30 minutes compared to 315 minutes for manual recovery – an efficiency index of 90.5%, with an automation coefficient Ka = 87.5%. It is demonstrated that with proper organization of persistent data storage at levels 3-5 of the proposed hierarchy, the Disaster Recovery Plan can be functionally replaced by an executable IaC repository, guaranteeing deterministic and reproducible recovery regardless of the availability of a qualified specialist.