Its more like this exsample
Client > Server cluster > Server hub EU >< Server hub NA < Sever cluster < Client
Instead of it going Client > Server cluster > Server NA
It lowers the rate of data from both servers via distance and traffic because what is happening is the main server hubs are taking there data from there respected areas and then the server hubs trade data to each other reducing the load and as i said distance and traffic of all the data problem with this method is can cost a bit and needs a lot of coding for the data transfer to behave itself.