In this paper, Big-Data Driven Digital Ecosystem Framework (BDDDEF) for Online Predictive Control Systems is created. The proposed framework consists of different Agents, where each Agent is a distributed and virtual service. In our work, we provide solutions to the Big Data challenges in building Digital Ecosystems for Online Control including high volumes, velocity and variety of data, and the need for low data latency. We propose to use BDDDEF for building robust, reliable, fault-tolerant, scalable and high-loaded data pipelines for Online Predictive Control Systems. We review Big Data Main Systems for Online Predictive Control Architecture, review the literature for Digital Ecosystems design for Control Systems Online, design and describe main features, main architectural components and functional architecture of the framework, and finally, propose new Predictive Control methodology for Online Predictions.