Background: High-load systems (HLS) underpin critical digital infrastructure ranging from financial transaction processing to cloud-native platforms, demanding continuous monitoring to maintain Service Level Agreement (SLA) compliance and end-user satisfaction. Despite widespread adoption of monitoring tools, the literature lacks a unified analytical framework that integrates queueing-theoretic models, reliability engineering, and evidence-based decision-making for real-time scalability management. Methods: This study addresses this gap by: (1) formalizing critical monitoring metrics through the lens of queueing theory and reliability mathematics; (2) proposing an anomaly detection procedure based on the Irwin criterion with empirical validation; (3) developing a logistic saturation model of system throughput calibrated against load-testing data across four operational scenarios; (4) constructing an Analytic Hierarchy Process (AHP)-based matrix for ranking SLA factors by consumer importance; and (5) articulating an evidence-based framework for scaling decisions. Experimental load testing of a microservice-based e-commerce platform (8-node Kubernetes cluster) was conducted. Results: Horizontal scaling from 2 to 16 instances reduces 99th-percentile latency by 73.2% while maintaining 99.97% availability under 10,000 concurrent users. The proposed logistic model predicts saturation onset within 4.1% of observed values. The integration of predictive monitoring, mathematical modeling, and structured evidence-based reasoning significantly enhances the capacity of HLS operators to anticipate failures, optimize resource allocation, and sustain SLA compliance under dynamic load conditions. Conclusion: The proposed framework demonstrates that combining queueing-theoretic modeling, anomaly detection, and structured evidence-based reasoning provides a robust approach to real-time scalability management in high-load systems. Limitations and directions for future research are discussed.
Stepanov et al. (Tue,) studied this question.