eBPF Model Discovery Experiments: Code, Analytical Results and Worked Example
Year: 2026
Type: software
Abstract: This research artifact accompanies a study of eBPF-assisted discovery of operation availability models when application-level SDK tracing is absent or incomplete. The artifact contains the study source code, experiment configurations, analytical results, provenance information, computational cost records, and one selected example containing real training observations and a frozen model. The corrected analysis comprises 108 cases: 72 original Petclinic and OpenTelemetry Demo cases and 36 corrected Social Network cases. Historical Social Network results are retained to document the correction and are not additional independent replications. The uploaded archive contains: code-and-results.zip: source code and analytical data; worked-example.zip: observations and a frozen model for the selected OpenTelemetry Demo case with SDK tracing absent and OBI collection; manifests, checksums, and build information. The source snapshot is commit 22bac95b046346e5a292776489eac3ecbf179451 of https://github.com/a-a-k/ebpf-model-discovery-experiments. The artifact supports inspection of the implementation, new executions using the study code, reaggregation of saved analytical results, and examination of the selected training/model example. It does not contain individual request latencies or complete classification assignments for every historical run. Consequently, historical latency quantiles and adjusted Rand index values cannot all be rederived from individual observations using this artifact alone. During package verification, correction statistics recomputed from the included analytical data matched the saved correction report. The packaging workflow did not execute new measurements or rebuild a model. Original source code authored for this study is licensed under the MIT License. Original research data and documentation are licensed under the Creative Commons Attribution 4.0 International License. Third-party materials retain their original licenses.
Author: Krasnovsky, Anatoly A.
Institution: Innopolis University
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225784434
Cites:
Cited by:
Related to: