Semantic grid resource monitoring and discovery with rule processing based on the time-series statistical data

This paper presents a new extension to a semantic grid resource monitoring and discovery system called S-MDS[1]. The extension requires S-MDS to store a sequence of resource property values into an RDF database so that aggregate and statistical calculation can be performed over the values. By using a rule-based approach combined with the inference capability of the RDF database, S-MDS provides a novel important function, namely, resource anomaly detection in OGSA-based grids. For example, it is possible to lively monitor and detect CPU load anomaly that exceeds 3 times of the standard deviation from the average of past two weeks loads. Finally, the extension gives great flexibility to users to define their own monitoring rules by using a general purpose rule language.