Are you using container managed persistence? If not, it will be application specific, I'm afraid.
If it's container managed, are you eventually getting connection timeout exceptions?
If not, the only possibility that I can think of at the moment is that you're somehow keeping connections active but not doing anything with them. The configuration solution for this (as opposed to the programming solution, which should likely be investigated depending on the impact of the outage) would be to use the "aged timeout" setting in the datasource's connection pool. In admin console, navigate as follows:
JDBC providers > [your db2 provider] > Data sources > [your datasource]> Connection pools
Reference here for more detail on this:
http://publib.boulder.ibm.com/infocenter/wasinfo/v6r1/index.jsp?topic=/com.ibm.websphere.base.doc/info/aes/ae/umj_sesspoolset.htm