Visualising distributed applications' performance
For many large-scale problems, such as earth observation, industrial and scientific modelling, as well as biomedicine, new approaches to computing are continuously required. The CROSSGRID project sought to address realistic problems in environmental protection, surgical planning and analysis of large volumes of data from the high-energy physics experiments conducted at CERN. Whilst existing solutions only enabled such Grid applications to be run in a batch mode, with CROSSGRID it is now possible to monitor the progress of the application's execution. More specifically, the concept of interactive Grid computing was supported by innovative products developed to permit changes in the application's parameters and ensure that end users receive output on the fly. G-PM, providing a basic measurement system, was created at the AGH University of Science and Technology in Krakow for monitoring grid systems through user-defined metrics. Prompted by the user when applying custom metrics, the server can perform asynchronous "probes" to read and write the state of the application and return measured data. Furthermore, custom metrics can be derived automatically from an existing set of metrics. The G-PM tool consists of three main components. The performance measurement component (PMC) provides the functionality for standard performance measurements of both Grid applications and the Grid environment. The aim of the high level analysis component (HLAC) is to provide application developers with application-specific data, based on a metrics specification language that can combine performance data from different sources. Finally, the user interface and visualisation component (UIVC) allows the user to specify performance measurements and visualise the performance data collected. This software is freely available at the project website, http://www.crossgrid.org/ under an open source license.
