Comparative Performance of Operating Systems in Handling Sentiment Analysis REST API Workloads
Year: 2026
Type: article
Abstract: This study evaluates the performance of a sentiment analysis REST API deployed across different operating system environments, with particular attention to tail latency and available operating-system-level execution indicators. The study examines how differences in platform-level execution conditions and scheduling-related behavior may be associated with variations in service performance under parallel workloads. Comparative experiments were conducted using Apache JMeter with incremental concurrency, while available system monitoring data were used to observe context switching and CPU utilization patterns. All recorded scenarios achieved a 0% error rate. In the available Windows Server and Ubuntu results, Ubuntu exhibited a lower P99 latency (179.40 ms) than Windows Server (416.53 ms), although Ubuntu experienced a single extreme maximum latency spike of 1792 ms. Ubuntu monitoring data also indicated uneven CPU utilization, with one core reaching 91.69%, suggesting a potential workload concentration issue. Windows Server exhibited a higher P99 latency together with context switching peaks of 2765 per second. The macOS results showed lower recorded latency values; however, because the available test conditions did not achieve equivalent workload parameters and synchronized monitoring across all platforms, these findings are interpreted descriptively rather than as evidence of superior operating-system or scheduler efficiency. Overall, the results demonstrate the value of interpreting application-level latency together with available operating-system-level indicators when evaluating REST API deployment performance. The findings further identify worker configuration, CPU affinity, and background process isolation as potential areas for future optimization and controlled experimental validation.
Source: JURNAL TEKNIK INFORMATIKA
Institution: Universitas Budi Luhur
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7222986197