OPTIMIZATION OF MICROSERVICE INFRASTRUCTURE FOR REAL-TIME SYSTEMS: BALANCING PERFORMANCE, RESILIENCE, AND COST

In the era of pervasive digital transformation, microservice architecture has become firmly established as a foundational model for building high-load distributed systems. At the same time, achieving a sustainable balance among performance, fault tolerance, and the economic efficiency of infrastructure still remains one of the central scientific and applied challenges. The present study examines the limits of applicability of current service orchestration approaches and substantiates the existence of a research gap expressed in the insufficient integration of predictive scaling mechanisms with DataOps pipelines. The research objective is to develop and experimentally validate a comprehensive adaptive model of multi-criteria optimization for cloud infrastructure oriented toward real-time systems. The methodological framework of the study is built upon a systematized review of scholarly sources, a comparative analytical approach, and an empirical case analysis of high-load environments in which the daily volume of processed data reaches 8 TB. The obtained results indicate that the use of hybrid auto-scaling mechanisms in combination with topologically sensitive routing ensures a reduction in data-processing latency by 88-93%, a decrease in mean time to recovery (MTTR) by 85%, and an increase in infrastructure availability to the level of 99.95%. The final conclusions confirm the achievement of the stated research objectives and the validity of the proposed model. The presented provisions possess substantial theoretical and practical significance for specialists engaged in the design and operation of resilient cloud-native solutions, including system architects, senior DataOps and DevOps engineers, as well as IT executives