This paper considers storage management in an isolated WSN, under the constraint that the storage space per node is limited. We formulate the memory spaces of these sensor nodes as a distributed storage system. Assuming that there is a sink in the WSN that will be visited by mobile mules intentionally (e.g., pre-arranged buses) or occasionally (e.g., non-pre-arranged taxis), we address three issues: (1) how to buffer sensory data to reduce data loss due to shortage of storage spaces, (2) if dropping of data is inevitable, how to avoid higher priority data from being dropped, and (3) how to keep higher priority data closer to the sink, such that the mobile mules can download more important data first when the downloading time is limited. We propose a Distributed Storage Management Strategy (DSMS) based on a novel shuffling mechanism similar to heap sort. It allows nodes to exchange sensory data with neighbors based on only local information. To the best of our knowledge, this is the first work addressing distributed and prioritized storing strategies for isolated WSNs.