Please provide an example of your slow queries. Try doing a count on the same query instead of extracting rows (count(*)) - is that a lot faster? If it is then it is likely that Impala to SAS network connection is the bottleneck. If there is no difference in performance then it is the Impala database that is slow and you would need to talk to a DBA to find out why.
Have you tried the query inside Impala? For such small queries, Hadoop is slow, at least slower than SAS.
Data never sleeps
The SAS Users Group for Administrators (SUGA) is open to all SAS administrators and architects who install, update, manage or maintain a SAS deployment.