In accordance a method for storing a dataset, the dataset may be split into fragments that are distributed among different nodes of a network for storage. The fragments may then be retrieved as and when needed and re-assembled. The method allows multiple different fragments to be stored and re-assembled on demand. The dataset is initially stored in a user computing device in communication with a data storage system and a custodial entity. The fragments are stored so that no single computing entity in the storage system or the custodian ever contains or gains knowledge of all the fragments. Additionally, the user computing device that was initially in possession of the dataset and which caused the fragments to be stored in the storage system may be replaced with different user computing devices without losing the capability of storing and re-assembling the user data on demand by the replacement user computing device.