Message-Centric Monitoring
Monitoring for large-scale, long-running applications provides useful information about the status of a running application. Traditional trace-based monitoring approach generates large volumn of tracing data that are difficult to understand on-line.
Message-Centric Monitoring uses existing communications inside application to propagate performance information, therefore has low overhead. The information that it gathers is limited, however can be useful for the user to grasp an overall status of the application, for example, how fast is the application progressing. Most importantly, the interested information is available on-line to all processes.
Publications
Wenbin Zhu, Patrick G. Bridges and Arthur B. Maccabe, “Embedded Gossip: Lightweight Online Measurement for Large-Scale Applications”, ICDCS 2007, to be published.
Wenbin Zhu, Patrick G. Bridges and Arthur B. Maccabe, “Online Critical Path Profiling for Parallel Applications”, Cluster 2005.
Patrick G. Bridges and Arthur B. Maccabe, “IMPuLSE: Integrated Monitoring and Profiling for Large-Scale Environments”, LCR 2004.
