How do you (or can you) reconnect a remote view that was based on a connection that is now broken?

As of SQL Server 2005, MSFT seems to have implemented a "KeepAlive" timeout with a 30 second default setting to try to detect orphaned connections. If a client machine, (for whatever the reason), fails to respond to these KeepAlive signals, SQL assumes the client connection to be orphaned and disconnects from it in order to recover SQL Server system resources. All that is well and good, except that KeepAlive protocol that I am familiar with is implemented in the TCP protocol and not at the application level. So if you have a remote view connection to SQL Server in your application, and SQL Server disconnects your client, it seems that you are stuck!

Short of trying to convince the SQL DBA to zero out or otherwise turn of the KeepAlive feature of SQL Server 2005, or argue that if they are going to use that feature they should ensure that all their client machines and network routers support it, what can be done about this condition? Is it possible to recover on the fly and reconnect your remote view by dynamically changing the connection handle?

Any thoughts or help would be appreciated.