In distributed storage networks, it is important to design a scalable data structure for efficient data retrieval. Generally speaking, tree structures and hash tables are common ways to solve this kind of problems. In this paper, we present a novel distributed lookup protocol, called Dynamic Hash, to address this problem. The proposed Dynamic Hash is a scalable structure which provides support for locating the data stored on a node and efficiently handles nodes joining and leaving the system. Besides, it can easily achieve load balancing by adding nodes and reduce power consumption by removing nodes dynamically. Compared with related hash schemes, the proposed Dynamic Hash protocol is more scalable, relieves hot spots and requires very little overhead.