Recent years we have witnessed a great increase in modeling data as large graphs in multiple domains, such as XML, the semantic web, social network. In these circumstances, researchers are interested in querying the large graph like that: Given a large graph G, and a query Q, we report all the matches of Q in G. Since subgraph isomorphism checking is proved to be NP-Complete, it is infeasible to scan the whole large graph for answers, especially when the query's size is also large. Hence, the "filter-verification" approach is widely adopted. In this approach, researchers first index the neighborhood of each vertex in the large graph, then filter vertexes , and finally perform subgraph matching algorithms. Previous techniques mainly focus on efficient matching algorithms, paying little attention to indexing techniques. However, appropriate indexing techniques could help improve the efficiency of query response by generating less candidates. In this paper we investigate indexing techniques on large graphs, and propose an index structure DSI(Distance Set Index) to capture the neighborhood of each vertex. Through our distance set index, more vertexes could be pruned, resulting in a much smaller search space. Then a subgraph matching algorithm is performed in the search space. We have applied our index structure to real datasets and synthetic datasets. Extensive experiments demonstrate the efficiency and effectiveness of our indexing technique.