When using the BerkleyDB store, it is possible to create a feature with a 
very high ID by calling 'create' on a feature with a large ID. Doing this 
will cause the BerkleyDB store to perform much more slowly. There is 
currently no known work around for this problem.