Exploring Patterns and Correlations in CMS Computing Operations Data with Big Data Analytics Techniques

The CMS experiment at the LHC accelerator at CERN designed and implemented a Computing model that allowed successful Computing operations in Run-1 (2009-2012) and gave a crucial contribution to the discovery of the Higgs boson by the ATLAS and CMS experiments. The workflow management and data management sectors of the model have been operated at full capacity exploiting WLCG resources for years. Around the massive volume of original and derived physics data from proton-proton and heavy-ions collisions in CMS, plenty of other data and metadata about the performances of the computing operations have been also collected and rarely (or never) examined. This latter sample is a wild mixture of non-physics heterogeneous data, both structured and unstructured, which well fits to deeper investigation with Big Data analytics approaches. In the context of CMS R&D activities, exploratory projects have been started to extract some values from this dataset and to seek for patterns, correlations as well as ways to simulate the Computing Model itself. Such studies will be presented and discussed.