[KIEKER-1002] Visualize software performance characteristics (e.g., in XY-plots)

[rju]

JIRA Issue: KIEKER-1002 Visualize software performance characteristics (e.g., in XY-plots)
Original Reporter: Jan Waller

---

Brief explanation:

In addition to its use for [architecture discovery and visualization|Kieker:architecture-discovery/] based on monitored traces, Kieker can be employed as an [Application Performance Monitoring (APM)|Kieker:apm/] tool. To enhance Kieker's APM support, visualization components for performance characteristics, such as the response times of method calls, are needed. One example of such a visualizations would be a simple XY-plot with the elapsed time and the respective measured response time as axes. Many further possibilities are imaginable (histograms, density plots, etc.). Some examples are available in the [Kieker live demo|Kieker:demo/]; however, they are not in a state that supports easy reuse.

The available TSLib ( KIEKER-830 Done ) provides support for the analysis of time series based performance data. Of further interest can be the development of (generic) filters to support grouping or aggregation of collected data for downstream visualizations (e.g., aggregated by trace, all entry methods, ...).

To summarize, the goal of this project idea is providing a set of generic and configurable filters for common visualizations of (monitoring) data.

With respect to the used technology, it would be good if the developed solution integrates nicely with our ​Web-based UI; hence, SVG in combination with JavaScript/Ajax would be a possible option.

For the evaluation of the developed solution, we have quite a large set of trace data from our industrial case studies (which could also be replayed using Kieker in the online trace visualization case).

Related tickets:

• KIEK-868: Live installation/server of instrumented system + analysis/visualization
• KIEK-793: Basic XY time series plotter
• KIEK-830: Integrate TSLib (Time Series Lib) into core distribution
• KIEK-1114: Integrate OPAD into core distribution
• KIEK-627: Feature wish: Flame graphs
• KIEK-1158: Feature wish: Integration with Graphite/Whisper for storing/displaying time series data
• KIEK-835: WebGUI Cockpit
• KIEK-1060: Interactive trace visualization (2D/3D)

Expected results:

After the successful completion of this project, Kieker provides mechanisms to visualize different performance characteristics collected in monitoring logs. The visualization may be generated offline or online.

Knowledge prerequisite:

The student working on this topic should

• be a practiced Java developer
• be interested in software performance engineering (SPE)
• be interested in visualizations of (performance) data
• have basic knowledge in the chosen technologies (can be acquired during the project)

Skill level:

• low to medium

Additional documentation:

• Some resources on the visualizations already supported by the TraceAnalysis tool:
  • The [Kieker user guide|Kieker:documentation/] documents the visualizations currently included in the Kieker release.
  • André's Kieker Days 2012 talk includes trace visualizations from different case study systems:
    • van Hoorn, A. (2012) Kieker: Overview, Review, and Outlook (Talk) In: KoSSE-Symposium Application Performance Management (Kieker Days 2012), November 29/30, 2012, Kiel, Germany.
  • Bettual Richter's and Felix Magedanz's theses include visualizations from a COBOL and a C# system:
    • Richter, B. (2012) Dynamic analysis of COBOL system architectures for model-based testing (Diploma thesis, in German), Kiel University, Kiel, Germany, 116 pp
    • Magedanz, F. (2011) Dynamic analysis of .NET applications for architecture-based model extraction and test generation (Diploma thesis), Kiel University, Kiel, Germany, 132 pp
• Visualization of performance anomaly detection results (combination architecture/data):
  • Marwede, N. S. (2008) Automatic Failure Diagnosis based on Timing Behavior Anomaly Correlation in Distributed Java Web Applications (Diploma thesis), University of Oldenburg, Oldenburg, Germany, 119 pp
• Time series based analysis and visualization:
  • Bielefeld, T. C. (2012) Online performance anomaly detection for large-scale software systems (Diploma thesis), Kiel University, Kiel, Germany, 133 pp

Candidate Mentor(s):

The primary contact for this project idea is marked in bold font.

• André van Hoorn ([Kieker developer|Kieker:author/avh/]; researcher with University of Stuttgart)
• Björn Weißenfels (student assistant with Kiel University)
• Florian Fittkau (researcher with Kiel University)
• Jan Waller ([Kieker developer|Kieker:author/jwa/]; researcher with Kiel University)