Barbie, Alexander (2016) Runtime Information Integration into System Dependency Graphs. (Student research project), Institut für Informatik, Kiel, 15 pp.

Preview

Text stu97569_final.pdf - Published Version Download (818kB) | Preview

Abstract

Within the last decade, the importance of multi-core processors
increased, due to a leak in performance improvement of single-core
processors. As a consequence, software engineers need knowledge about
concurrency issues. They must be qualified to meet performance requirements
and to find bugs in concurrent programs.
We enhance a semi-automatic, pattern based approach to support software
engineers in the parallelization process of sequential Java programs,
by adding runtime information to the system dependence graph in a
graph database. Therefore, we use an extension of Soot to find system
dependencies and save them in a Neo4J graph database. The runtime
information is gathered by Kieker. Further, we propose a tool, based on
a pipe and filter architecture realized with TeeTime, to add these runtime
information via Cypher queries to the Neo4J database. All gathered
information can be visualized via the browser. This tool adds runtime
information correctly to the graph database. However, the tool’s correctness
depends on the applications we utilize. If the system dependence
graph contains errors, a possible relationship will not get the supposed
runtime information.

Actions (login required)

View Item