In a paradigm of value creation from data collection and analysis, using Big Data, and particularly mining techniques, industrial actors are confronted to a series of issue regarding data collection, storage, distribution, and management. In this paper, we worked with several French industrials to provide them a data platform, able to centralise and distribute data coming from the shop-floor and from business related information systems. Our experience with both green field and brown field approach, as well as an historic activity in big data platforms for telecommunication industries have been used to compile a series of definition, working concepts and recommendations regarding the integration of a data platform to support manufacturing, data-based applications. These lessons learnt are presented in this paper, as well as on-going experiments to further improve key components and functions of industrial data platforms.