2015 · IEEE BigData Congress

A Flexible Data-Driven Approach for Execution Trace Filtering

Kadjo Gwandy Kouame | Naser Ezzati-Jivan | Michel R. Dagenais

Evidence basis: full-text-reviewed · Review status: catalog-reviewed; paper-author approval pending

kernel-tracing system-tracing trace-filtering trace-abstraction trace-analysis

execution trace filtering LTTng 2.4.0 Trace Compass finite-state machine declarative analysis synthetic events SYN flood hping3 state system XML modeling kernel tracing

Core contribution: The paper provides a declarative XML/state-machine approach for flexible filtering and synthetic-event generation over large kernel traces.

Problem and motivation

Large execution traces contain many irrelevant events, while hard-coded filters cannot flexibly express changing event combinations, temporal constraints, or late-arriving events (pp. 2-3).

Method and contribution

Users define XML FSMs with transition inputs and actions. XML is converted into Java models and executed in a Trace Compass plug-in over a generic state-system/interval-tree container. Inputs may be raw/synthetic events or event/time conditions; actions can change state, emit synthetic events, start FSMs, or trigger UI selections (pp. 3-5).

Findings and evidence

The SYN-flood example uses two coexisting FSMs, a 3-second timeout, and a 100-attempt threshold. hping3 generates the workload; LTTng 2.4.0 records inet and kernel events on Ubuntu 14.10/Linux 3.13.0-43 with Apache. The trace yields 569 half-open-connection synthetic events and one SYN-flood event. For a 150 MB trace, XML model construction averages 70.092 s versus 67.663 s for a hard-coded Java model (pp. 6-7, Table I).

Limitations and future directions

Limitations: Runtime grows with active scenarios and condition checks; the example can process about 1,707 conditions per event. Complex FSMs are harder to understand/debug, and the paper says comprehensive performance evaluation was not yet possible (p. 7).

Future work: Add more patterns, optimize execution, and refine the XML language and user-facing analysis support (p. 7, conclusion).

Sources and identifiers

When to cite this paper

Cite this paper when your work uses or compares declarative XML FSMs for user-defined trace filtering and synthetic-event generation.

Citation

BibTeX
@inproceedings{ezzatiJivan2015aflexible,
  author = {Kadjo Gwandy Kouame and Naser Ezzati-Jivan and Michel R. Dagenais},
  title = {A Flexible Data-Driven Approach for Execution Trace Filtering},
  year = {2015},
  booktitle = {IEEE BigData Congress},
  pages = {698-703},
  publisher = {IEEE},
  doi = {10.1109/BigDataCongress.2015.112},
  url = {https://doi.org/10.1109/BigDataCongress.2015.112}
}
Other citation formats for Word and reference managers
APA 7
Kouame, K. G., Ezzati-Jivan, N., & Dagenais, M. R. (2015). A Flexible Data-Driven Approach for Execution Trace Filtering. In IEEE BigData Congress (pp. 698-703). https://doi.org/10.1109/BigDataCongress.2015.112
IEEE
K. G. Kouame, N. Ezzati-Jivan, and M. R. Dagenais, "A Flexible Data-Driven Approach for Execution Trace Filtering," in IEEE BigData Congress, pp. 698-703, 2015, doi: 10.1109/BigDataCongress.2015.112

Readable Markdown record · JSON record · Download RIS