Architectural Patterns for Designing Fault-Tolerant Microservice Systems within a Platform

The article examines architectural patterns for designing fault-tolerant microservice systems within a platform in the context of increasing distribution, workload intensity, and requirements for continuous availability in corporate and industrial environments. The study integrates findings describing platform-level decomposition, resilient models of inter-service interaction, asynchronous communication, replication, orchestration, and automated recovery mechanisms, as well as the impact of service mesh technologies, self-adaptive patterns, and monitoring loops on system behavior under partial failures. It is shown that fault tolerance emerges not as a property of individual services, but as an effect of the coordinated organization of the server and communication layers, where traffic concentration and control nodes, as well as the message-transport layer, become critical. A distinction is substantiated between architectural indicators of resilience and application-level metrics of computational tasks, since changes in model quality and data characteristics may not reflect degradation in platform connectivity. Based on a comparative analysis of approaches, a conceptual framework is proposed that describes a set of resilient patterns and the boundaries of applicability of empirical indicators depending on workload profiles, infrastructure configuration, and failure scenarios. The article may be useful for architects, engineers, and decision-makers designing microservice platforms with requirements for scalability, predictable behavior, and continuity of computational processes.