I need some help understanding how hub transport is working in my environment and how to get it to work better.
First, an explanation of our environment.
We have about 110 branch offices, all connected to an MPLS network - meaning that all can see each other. We have two production datacenters connected to the MPLS network with a 45Mb pipe. Each datacenter has an Exchange 2007 server with both Mailbox and Hub Transport roles.
Two of the branch offices are large and have their own Exchange server with mailbox and transport roles installed. These are connected to the MPLS network with @10Mb pipes. The remaining branch offices connect to the MPLS network with T-1's.
We did not setup AD sites as we expanded, therefore everyone is in the same site. Let me add that my understanding of sites is very limited.
I have found that our datacenter Exchange servers are heavily taxed. We use Citrix in about half of our branch offices, so users will often get messages in Outlook reporting that it is waiting for the Exchange server. I believe that if we separate the transport role to its own server, that we will reduce the load on the mailbox servers.
I am concerned, however, with unnecessary mail traffic traversing the network. If I have a message in Site A destined for a mailbox in Site B, I don't want to use the transport server in Site C. Especially since the two Exchange servers are connected via heavily taxed 9-12 Mb pipes.
So the thought was to create new Transport servers in the datacenters. This will clearly relieve the load on the mailbox servers. But what happens to mailflow? Would it be better to have the transport role in one datacenter only and mailbox servers elsewhere? How would mail flow in this scenario?
Thanks in advance for your help. You guys are great.