Data and Analysis Scripts for: Impact of OpenTelemetry and ELAS Adaptive Sampling on Node.js Application Performance

This dataset contains the raw experimental results and R analysis scripts from a study evaluating the performance impact of OpenTelemetry instrumentation and the ELAS adaptive sampling middleware on a Node.js application deployed on Kubernetes. The experiment was conducted under controlled load using JMeter, with four concurrent user levels (150, 300, 600, 1200) and nine replicates per configuration. Two datasets are provided: one with Node.js Event Loop monitoring enabled (raw data) and one without (cleaned data, with outliers and warm-up runs removed). Each run includes JMeter results, pod-level CPU and memory metrics, runtime instrumentation metadata, and ELAS controller state snapshots.The R scripts perform paired statistical inference using the OOAP methodology (Hodges-Lehmann estimator, 95% confidence intervals, rank-biserial correlation, Benjamini-Hochberg correction) to compare instrumentation profiles across throughput, latency, and resource utilization metrics. The ELAS middleware source code (TypeScript) is also included, implementing a JMeter latency-based adaptive sampling policy that dynamically adjusts the tracing sampling rate under increasing system load. Together, these files enable full reproduction of the article's figures, tables, and statistical analyses.