This poster was presented at Computer Graphics & Visual Computing (CGVC24) at City St George's, University of London.
Kaleem Peeroo, Vladimir Stankovic, Peter Popov, Tillman Weyde
Data Distribution Service (DDS) is a specification for middleware in real-time and mission-critical systems such as energy management systems, air traffic control, and medical devices. Such systems have specific performance requirements to function properly and avoid catastrophe. However, the performance of the system can vary depending on the various parameters defined before or during run-time. Understanding how the different parameters can impact performance helps DDS system designers in setting up distributed systems in critical domains. This paper introduces a workflow for understanding how parameter values can affect system performance as seen in their distribution form. This workflow applies not only to DDS but any configurable system that has multiple parameters that can take multiple values that may result in different system performance. This visualisation workflow is demonstrated on a dataset created by running tests on actual DDS. We argue that they help understand non-linear relationships between the parameters and DDS performance and help identify anomalous results that may need investigation.