Transitioning to Active Directory Site-Based Routing

 

By Kate Follis

Are you still waiting to make the transition from Microsoft Exchange Server 2003 or Exchange 2000 Server to Exchange Server 2007? You're really going to like all the commands available in the Exchange Management Shell - that's a pretty cool feature and one that even administrators who are hooked on wizards can quickly learn. And you're probably pretty excited about implementing server-based rules and all the anti-spam improvements. But maybe you're hesitant about the idea of collapsing all your routing groups and maintaining mail flow while you systematically upgrade your Exchange organization. No problem. Just let me explain!

The New Exchange 2007 Routing Topology

One of the biggest changes between earlier versions of Exchange Server and Exchange 2007 is the move to a routing topology that is based on Active Directory directory service sites and IP site links instead of on routing groups and routing group connectors. Some Exchange administrators may feel a loss of control over their routing topology, but there's no need to worry. And there are plenty of reasons to celebrate. Exchange 2007 gives you the tools that you need to tweak the Active Directory routing behavior when it doesn't conform to the way that you want Exchange mail to flow. Plus you benefit from the improved network utilization of Active Directory site-based routing and no longer having to maintain a routing topology.

In our Exchange 2007 planning guidance (Best Practices for Transitioning an Exchange Organization), we make a point to show you all the pieces of the current topology that you have to document. This includes the configuration of Active Directory sites and IP site links, the location of domain controllers and global catalog servers, the configuration of existing routing groups and routing group connectors, and the location of currently deployed Exchange servers. I know: That's a lot of information to digest.

Fortunately, the Microsoft Exchange Best Practices Analyzer can take a peek at your network for you and let you know if the current topology will work for Exchange 2007. It also recommends improvements that you can make before the first Exchange 2007 server is even deployed. For more information about how to run the Exchange Best Practices Analyzer Exchange 2007 Readiness check scan, see Best Practices for Transitioning an Exchange Organization. And for more information about how to plan your deployment, see Planning Your Deployment. (All these references are in the Exchange 2007 Help documentation.)

By the way, when it comes to coexisting with Exchange 2007, there are no differences between the way that Exchange 2003 and Exchange 2000 behave. So, unless the documentation states otherwise, you can assume that everything that applies to Exchange 2003 also applies to Exchange 2000.

Making the Transition

If you are managing a single Exchange server or your Active Directory environment has just one site, your worries are over. When you deploy the first Exchange 2007 server that has the Hub Transport server role installed, a routing group connector is established between the servers that are running different versions of Exchange Server so mail continues to flow while you move mailboxes and transfer any connectors to external domains from servers that are running earlier versions of Exchange Server to Exchange 2007. If your Exchange organization is running multiple servers, you can modify the default routing group connector to add source servers and target servers. Making those configuration changes provides fault tolerance and load balancing.

If your Active Directory forest contains multiple sites, make sure that you take some time to plan your transition. Because two different administrators may manage Exchange Server and Active Directory in your organization, you might not be sure that the topology created by the Active Directory administrator is suitable for Exchange Server. However, the reasons for defining Active Directory sites are the very same reasons that you defined Exchange routing groups! Both are engineered to optimize traffic over the underlying network. You may be pleasantly surprised to discover that your current routing group configuration actually mirrors the Active Directory site configuration. Chances are that your routing group connectors are also following a similar path as the IP site links.

You can find out all about the Active Directory site and IP site link configuration by using commands in the Exchange Management Shell. For example, to see a list of all the Active Directory sites in the forest, type the following command:

Get-AdSite | format-list

To see a list of all the IP site links configured in the forest together with the Active Directory sites with which they are associated, run this cmdlet:

Get-AdSiteLink

For more information about the Exchange Management Shell, see Using the Exchange Management Shell.

There are some big differences between the way that Exchange 2003 and Exchange 2007 perform message routing, and also there are some similarities. Think for a minute about why you created routing groups and routing group connectors. When you created these objects, you defined how Exchange 2003 servers communicate. Exchange 2003 servers in the same routing group can communicate directly with one another. Exchange 2003 servers in different routing groups have to use routing group connectors to communicate, and each routing group connector defines a point-to-point connection between routing groups. When no point-to-point connection between routing groups exists, SMTP traffic has to relay from routing group to routing group to get where it is going.

In Exchange 2007, Active Directory site membership determines which Exchange 2007 servers can communicate directly with one another. Exchange 2007 Hub Transport servers use the intra-organizational Send connector (implicit and invisible!) to relay messages to other Hub Transport servers, whether they are located in the local Active Directory site or in a remote Active Directory site.

Every time that a message leaves the confines of the routing group or Active Directory site where it originates, it must select a route to its destination. The algorithms used to determine routing paths in Exchange 2003 and Exchange 2007 are very similar. The server versions simply consider different configuration objects.

For more information about how Exchange 2007 uses site membership for routing messages and server discovery, see Planning to Use Active Directory Sites for Routing Mail.

Let's Take a Trip

The following analogy demonstrates the efficiency of the Exchange 2007 routing system compared to the Exchange 2003 routing system.

Imagine that your entire Exchange organization is a country with a transportation system based on toll roads. Because every road has a toll, you always plan the cheapest route. In Exchange 2003, each routing group represents a city in that country, and the routing group connectors are the roads between the cities. In Exchange 2003, message routing is like driving to your destination and stopping at every city along the way. If at some point during the journey, you encounter an obstruction, like a bridge that's out, road construction, or a road crew taking a long break, you have to stop to ask directions and hope that a suitable detour is available. What's more, the server that gives you the new directions must broadcast the detour route country-wide. All the maps must be updated to reflect the new route, and when the obstruction is removed, the maps are updated again. In the meantime, the detour can take you far from your destination.

Exchange 2007 doesn't work that way. In Exchange 2007, the Active Directory sites represent the cities and the IP site links are the roads, but you get to fly to where you're going, instead of drive. Once you know your destination, you hop on an airplane and fly over all the cities along the way. If for some reason you can't land at your destination city because of heavy fog, a blizzard, or baggage handlers on strike, the plane simply changes direction and lands in the closest possible city to your destination. Then it waits for conditions at your destination to change and resumes the flight when conditions allow. The flight path parallels the roadways, and each road still imposes a toll, so you select the cheapest route before you embark on the journey. This helps determine which city your flight diverts to if there are problems.

If there isn't a direct flight path between two given locations, you can specify that some or all flights must stop in a hub city before the journey can continue. When more than one person is on your flight, just like when a message is addressed to more than one recipient, and each person is going to a different destination, you all travel together until you reach a fork in the individual routes to your destinations. Then there's a short layover and passengers change planes as needed to reach their own destination.

Even better, if you don't agree with the tolls being charged to travel the roads between cities (i.e., the costs assigned to IP site links by the Active Directory administrator), you can change the rate that is charged to Exchange 2007 servers. Adjusting the tolls can change which flight path is the cheapest and may determine whether your flight is routed through a hub.

In Exchange 2007, these "transportation" efficiencies benefit your organization in the following ways:

  • Sticking with a single, deterministic route even when failures are encountered eases troubleshooting. You can easily determine the point in the routing path where problems occurred and begin investigations there. That's not so easy to do when messages take a detour route.

  • Fewer servers handle messages. Exchange 2007 takes advantage of the underlying IP network to eliminate interim hops between source servers and target servers.

  • The server-to-server communication overhead that exists in Exchange 2003 to keep detour routes up to date is eliminated.

Routing When Exchange Versions Coexist

Table 1 compares the way different versions of Exchange Server handle routing.

Table 1   Routing differences between versions of Exchange Server

Exchange 2007 Exchange 2000 and Exchange 2003

Exchange uses Active Directory sites to determine an intra-organizational routing topology. All Exchange 2007 servers are associated with a single routing group for the purposes of routing to earlier versions of Exchange Server.

Exchange uses routing groups to determine an intra-organizational routing topology.

Exchange determines the least cost routing path between Hub Transport servers by using Active Directory IP site link costs.

Exchange determines the least cost routing path between bridgehead servers by using routing group connector costs.

Exchange uses direct relay to deliver messages between Hub Transport servers.

Exchange relays through bridgehead servers in each routing group in the routing path.

When Exchange can't connect, it uses the least cost routing path information to back off from the destination until a connection can be made to a Hub Transport server. Messages queue at the reachable site that is closest to the destination. This behavior is known as queue at point of failure.

When Exchange can't connect to the next hop in a routing path, it tries to reroute the message over an alternative path.

When a message is being sent to multiple recipients, Exchange delays message splitting until a fork in the routing path is reached. This behavior is known as delayed fan-out.

When a message is being sent to multiple recipients, message splitting occurs immediately after recipient resolution.

Each Hub Transport server queries Active Directory separately to retrieve the routing configuration used to calculate routing tables and to receive configuration updates.

Exchange uses a link state table to store a routing table and advertises configuration changes by using link state updates. The routing group master retrieves updates from Active Directory and coordinates the propagation of link state changes that are learned by servers in its routing group.

Active Directory site-based routing is a great stand-alone approach when you have only Exchange 2007 servers in your organization. However, as long as Exchange 2003 coexists with Exchange 2007, the Exchange 2003 servers still have to deliver messages to Exchange 2003 mailboxes. This means that you can't get rid of the routing groups and routing group connectors until all mailboxes have been moved to Exchange 2007. If you have both Exchange 2003 and Exchange 2007 deployed in the same organization, you have to maintain your routing groups and routing group connectors during the transition phase. To get back to our transportation analogy, Exchange 2007 uses a routing group connector to transfer messages into the Exchange 2003 roadway system, and Exchange 2003 uses a reciprocal connector to transfer messages to the Exchange 2007 airway system. So, depending on the origination and destination cities, the journey may involve both driving and flying.

Exchange 2003 and Exchange 2007 also differ in how they "see" one another. Exchange 2003 doesn't know anything about the system of sites and site links that make up the Exchange 2007 routing system. Exchange 2003 sees all Exchange 2007 servers as belonging to one big routing group. As far as Exchange 2003 is concerned, the Exchange 2007 server next door has the same proximity as the Exchange 2007 server on the other side of the world.

Exchange 2007, on the other hand, understands both routing systems and is well-aware of the routing groups and routing group connectors used by Exchange 2003. You may have to create additional routing group connectors between the Exchange 2007 routing group and any Exchange 2003 routing groups to make sure that you don't get stuck taking the "scenic route" and take advantage of the most expedient route.

Table 2 compares message routing in a variety of Exchange 2007 coexistence scenarios. Let's take a look.

Table 2   Comparison of routing behavior between Exchange 2003 and Exchange 2007

Source server Target server Routing behavior

Exchange 2003

Exchange 2003

Only routing group connectors are considered to determine a least cost routing path. Messages are relayed by bridgehead servers in each interim routing group on the routing path.

Exchange 2007

Exchange 2007

Only Active Directory IP site links are considered to determine a least cost routing path. Messages are relayed directly from a Hub Transport server in the source Active Directory site to a Hub Transport server in the target Active Directory site.

Exchange 2003

Exchange 2007

Exchange 2003 determines the least cost routing path across routing group connectors to the Exchange 2007 routing group. When a message arrives in the Exchange 2007 routing group, the receiving Hub Transport server either delivers the message within the same Active Directory site or routes the message to its destination by using the least cost routing path across IP site links.

Exchange 2007

Exchange 2003

Exchange 2007 Hub Transport servers discover the least cost routing path across IP site links to reach a routing group connector that gets the message as close to its destination as possible. The receiving Exchange 2003 bridgehead server either delivers the message or routes the message to the next hop routing group connector until it reaches the destination.

For More Information

This article gives you a quick look at how to transition from routing groups to Active Directory sites. You can read a lot more about how to plan to use Active Directory sites for routing mail and coexisting with Exchange 2003 in the Exchange 2007 Help documentation. To get started, see the following topics:

Also, for a detailed discussion of the transition process, see "Upgrading Your Infrastructure to Exchange 2007" in the December 2006 issue of TechNet magazine.

Enjoy your flight!

a2a2b2cd-23e4-4ae8-99b6-1eb3cb5d522a Kate Follis - Senior Technical Writer, Microsoft Exchange Server