A data processing apparatus is deployed at a cloud network, and includes a plurality of data nodes, a receiver circuit, and an aggregation circuit. The receiver circuit is coupled to the plurality of data nodes and is configured to receive downtime records from a plurality of industrial machines, and to selectively route the downtime records to selected ones of the plurality of data nodes based upon a predetermined criteria. Each of the plurality of data nodes comprises a control circuit and a memory. Each control circuit is configured to, in parallel with the other control circuits, to further populate the downtime records with other data related to the operation of the machines, determine one or more time windows for each of the records, and divide the records according to the time windows.