Resource Management Method for Self-healing Critical it Infrastructure in a Multidomain Environment

Abstract

This article solves the scientific and practical problem of ensuring the reliability and quality of information service provisioning in distributed computing environments by developing a resource management method for self-healing critical IT infrastructure. The functioning processes of the IT infrastructure management system), which structurally and logically acts as the customer's internal IT department, are investigated. An approach to categorizing information services and formalizing dynamic Service-Resource Models is proposed, allowing the synthesis of independent duplicate structures for critical category services across multiple domains. The autoscaling and weighted load balancing algorithms are improved by applying the determination of the Pareto-optimal solution set and using the minimum normalized Euclidean distance to the ideal point within the solution set. The Pareto multicriteria optimization method is substantiated, based on mathematically reducing the deviation of actual reliability and response time metrics from their target and normative values to zero. The architectural scheme of the management system is formed, strictly separating the processes of service quality evaluation by monitoring tools and quality level management by the analytical module. Experimental modeling confirmed the ability of the developed system to isolate resource failures and adapt to the nonlinear growth of requests, performing component migration and topology reconfiguration without system downtime.