API тестування та перцентильний аналіз його продуктивності

Abstract

This study examines methodological aspects of API testing with an emphasis on percentile-based performance evaluation. Unlike mean latency metrics, percentile analysis provides a more representative view of response-time distribution under variable workloads. The paper justifies the use of P95 and P99 indicators for detecting performance degradation, improving SLA formulation, and enhancing reliability assessment in distributed systems.