Potentially not. The Swarm cluster will automatically redistribute replicas to different nodes, if it is cheaper overall to store them there. For example, if you store a stream on a node that is relatively full, with respect to disk space, and then a new empty node joins the cluster later, it may be the case that your stored stream is moved to the new node, which offers more efficient storage of the data. There are many other reasons content may be moved around the cluster as well. In general, any node in the cluster can find your stream given its UUID, but you should make no assumptions about which node stores it at any given time.