Dealing with Oracle over high-latency links

We have stand-by application-servers in multiple datacenters -- in case we must fail-over in a hurry -- and would like to make _some_ use of them during the times of normalcy. The Oracle databases are replicated from the master-instances located in the currently-active datacenter to the slaves located in the redundant ones. The application -- provided by a vendor -- can only talk to _one_ database, so we cannot even direct read-only queries to the Oracle-replica closer to it. Thus, all of the remote application-servers have to talk to the master Oracle-instance, that's geographically far from them. The latency is substantial: tens of milliseconds instead of a fraction of one, when the app- and db-servers are on the same LAN. This slows down some of the operations _tremendously_ -- 20-minute jobs turn into hours-long ones... The vendor frowns at it and says, there is nothing to do -- they never tested their software in such a configuration, and make no effort in batching up requests, for example. The app is "chatty", talking to the DB all the time and awaiting responses before sending new ones. And, maybe, there is nothing to be done, but I'm wondering, if we can tune some settings in the tnsnames.ora, that may help ease the problem -- even if by a little bit... Can we?