A data profile request is handles by utilizing data in a distributed file system. Tabular data is extracted from a data source and stored in a distributed file system. Each table in the tabular data is split by columns, which are each stored in separate files in a set of physical nodes of the distributed file system. In response to a data profiling request, a master node determines, based on the profiling request, which groups of files are needed to be on a same physical node in order to perform the profiling analysis. The master node creates jobs using physical nodes that contain the requisite files needed for each job.