Patent attributes
A computer-implemented method for performing an analytic workflow process is provided. The method includes storing data from multiple input sources into a data storage pool, performing at least one analytic workflow executed by computing nodes, identifying data, by an analyzer module, expected to be required for a downstream analytic step. Thereby, the identified data have not yet been received in the in-memory database. The method also includes triggering a push module to move the identified data from the data storage pool to the in-memory database. Additionally, the method includes monitoring, by the analyzer module, output data of an analytic step and identifying pointers to additional data expected to be required for a downstream analytic step of a parallel analytic workflow process different to the analytic step of the monitored output data, and moving, by a push module, the data referred to by the pointers to the in-memory database.
