I have a read-mostly application and have elected to use NamedQueries in place of hibernate DAO's.

I execute the following in Oracle, with trace on and the analysis indicates that it completes within 00:00:00 seconds,
it is after all a simple row count:

SELECT count(*) as totalRows
FROM (
SELECT A.KEY Key
FROM
VALUES A
WHERE A.action_date between to_date('9/1/2008', 'MM/DD/YYYY') and to_date('11/30/2008', 'MM/DD/YYYY')
) A

Consistent with the handling of more complex queries, when I save this as a named query and execute it as follows, the execution time takes just over a minute:

Query q = hibernateSession.getNamedQuery("GetCount");
q.setDate("fromDate", rqst.getFromDate());
q.setDate("toDate", rqst.getToDate());
q.setResultTransformer(Transformers.aliasToBean(CountDTO.class));
CountDTO c = (CountDTO) q.uniqueResult();

I am not using any caching strategy. Why would such a simple query take so long?
I wonder if my larger queries are suffering additional performance slow downs.

The Actual Named Query definition is as follows:
<sql-query name="GetCount">
<return-scalar column="totalRows" type="java.lang.Integer"/>
<![CDATA[
SELECT count(*) as totalRows
FROM (
SELECT A.KEY Key
FROM
VALUES A
WHERE A.action_date between :fromDate and :toDate
) A
]]>
</sql-query>

By rewriting the query to use the SQL Sum and general rearrangement of the sql, the performance was
improved. It wasn't apparently hibernate or named queries, so much as its execution of the query.