# Always completely consume the ResultSet or Iterator returned by a query
<a name="best-practices-gremlin-java-resultset"></a>

The client object should always completely consume the `ResultSet` (in the case of string-based submission), or the iterator returned by `GraphTraversal`. If the query results are not completely consumed, the server holds onto them, waiting for the client to finish consuming them.

If your application only needs a partial set of results, you can use a `limit(X)` step with your query to restrict the number of results that the server generates.