Toward industry 4.0, modern manufacturing companies are aiming at building digital twins to manage physical assets, processes, people, and places. Since in this environment, massive amounts of data have been generated and collected, integration and management of various data sources is of paramount importance. In this context, cloud computing as a crucial part of Industry 4.0 facilitates distribution of computer resources without direct active management by users. Accordingly, an ontology enables efficient integration and management of data as a reference data model through representation of knowledge. Besides, data mining from massive data is very important to identify significant meaning of data, and to avoid unexpected errors through predictions from experiences described in data replica. This research addresses problems towards efficient integration and management of data for predictive maintenance.