Weißenfels, Björn (2014) Evaluation of Trace Reduction Techniques for Online Trace Visualization. (Master thesis), Kiel University, Kiel, Germany, 82 pp.

Preview

Text bjw_msc_thesis.pdf - Published Version Download (1MB) | Preview

Abstract

Software landscapes are becoming increasingly complex. Knowledge about the communication
between the applications in the software landscape is often lost. The comprehension
of such software systems can usually not be solved by static code analysis. The use of
dynamic analysis in this process, e.g., with Kieker, leads rapidly to a large amount of trace
data. For several purposes like storage or live trace visualization, e.g., with ExplorViz,
these data must be reduced.
This thesis addresses this issue by presenting 12 different trace reduction techniques.
These techniques are based on summarization, metric-based filtering, or removing utility
methods. We evaluate all techniques based on our assessment criteria, and implement the
most promising techniques as filters for ExplorViz. We also experiment with 6 execution
traces of different object-oriented systems and study the gain attained by these reduction
techniques.

Actions (login required)

View Item