[ephimetheus]

For experiment data, there is a layer on top of all of this that distributes datasets across the computing grid. That system has a way to handle replicate at the dataset level.