Unified Availability Model (UAM): Methodology for Calculating the Availability of Heterogeneous IT Systems

The proposed Unified Availability Model (UAM) establishes a formal and extensible methodological framework for assessing the availability of heterogeneous information systems that differ substantially in their architectural properties, operating modes, and types of observable metrics. Unlike traditional SRE- and API-centric approaches that assume metric homogeneity and rely primarily on “latency–error” pairs, the UAM methodology introduces a universal normalization procedure that enables different metric types to be mapped to a single, comparable scale.

The model is based on four fundamental constructs—statistical baseline, SLA thresholds, physical system limits, and permissible deviations—each of which is formalized as an independent normalization operator. A weighted combination of these operators provides a uniform availability calculation for a broad range of systems: REST/gRPC APIs, payment gateways, cryptographic HSM modules, batch/ETL processes, ERP platforms (Oracle EBS, SAP, 1C), as well as EDI/IDoc integration channels.

At the top level, UAM introduces the concept of a business contour—a structural composition of heterogeneous subsystems—and defines contour availability as a weighted aggregated function of the availabilities of individual components. This makes it possible to obtain a single, interpretable indicator of the operability of an end-to-end business process, compatible with practical observability tools (Prometheus, Zabbix, Grafana, VictoriaMetrics).

The presented model offers both theoretical novelty—through the formalization of inter-system metric normalization—and high practical applicability, providing engineering teams with a unified approach to measuring, comparing, and managing the availability of complex corporate IT landscapes. UAM forms the basis for the development of a generalized mathematical theory for assessing the reliability of multilayer enterprise systems.