Long running WCF services timeout in Azure after one minute

While working on an issue today with one of my customers, we discovered an odd exception that was being thrown when his WCF service was deployed to Azure. The project was essentially a pub/sub service that used long=polling.  The customer set all of his timeouts to 5 minutes on both the server and the client.  The service was tested and it worked perfectly in the development fabric.  Then the WCF service (Server) was pushed to Azure and the client application (running locally) started to throw a CommunicationException after about a minute.  It turns out that the load balancers in Windows Azure terminate idle connnections after one minute.

There are no settings to adjust this.  You will need to adjust your 'Direct-Push' types of appplications accordingly.