What is DAG? (part 2)

I took a while Today to figure out how the DAG really work behind. And I will not walk through the whole steps of building a DAG, as you can find many well written instructions by Google. What I would like to talk about are something that help you understand DAG better.

Like my previous post, I also use my project as an example. Illustrated in the below diagram, you can see each mail server has two network interfaces. One NIC is in the MAPI network where the inbound/outbound mail traffics are. One NIC is in the dedicated replication network for database synchronization. In the EMC, under the DAG network, you can enable/disable the replication for each NIC. It is also OK if you use only one NIC for both. Two NICs are just recomended.

Once the DAG has been created, you have to assign IP addresses to it. One IP addres for each subnet. Only in this way can the DAG talk to each mail server in every subnet. In my opinion, DAG is something like a virtual device. As after you assigned the IP address to it, you can find the DAG name under the Computer container in AD.

Each mail server that’s a member of a DAG is also a node in the underlying cluster used by the DAG. You can find the cluster in the ‘Cluster Core Resources’ in the ‘Failover Cluster Manager’, the name is the same as the DAG’s.

Exchange 2010 includes a new component called Active Manager that replaces the failover managerment in the old version. Simply speaking, the Active Manager will trigger the failover process and decide which database will be used when a failure is detected.

Microsoft has a clear explanation on how the Active Manager works: http://technet.microsoft.com/en-us/library/dd776123.aspx