Unstructured manufacturing big data silos are challenging for enabling various data-driven applications such as digital threads and digital twins in manufacturing. The management of big data silos requires to address the issues of large volume, data inconsistency, data redundancy, information silos and data security. This research developed a systematic approach to managing data silos using the state of art big data software. Applying this approach in the product life cycle can control data silos, data consistency, redundancy, timely update and enable the automatic workflow of each system.