Defining Routing Groups

  • Article

 

As a general guideline, you should define one routing group and add others only when necessary. The fewer routing groups in your environment, the less complex and more manageable it is. However, geographical and administrative requirements, as well as network availability, may mandate the creation of additional routing groups.

Routing groups are generally created for one of two reasons:

  • To accommodate varying network connectivity across servers.

  • To restrict the usage of a connector to users in a particular area. For more information about using routing groups to restrict connector use, see Understanding Connector Scope and Restrictions.

Before you define your routing groups, consider the advantages and disadvantages of multiple routing groups as shown in the following table.

Advantages and disadvantages of multiple routing groups

Advantages of multiple routing groups Disadvantages of multiple routing groups

  • Allows scheduling and control of mail flow. You can restrict connector use to a particular routing group or schedule the use of a connector.

  • Allows you to control usage based on message size or content by using connector restrictions.

  • Introduces more hops en route to the final destination, thereby decreasing delivery efficiency.

  • Adds complexity to your messaging environment.

  • Can reduce the reliability of messaging because the more hops you have en route, the more points of failure are possible.

  • Simple Mail Transfer Protocol (SMTP) handles latency in a well-connected TCP/IP environment, and this often eliminates the need for multiple routing groups.

  • Two routes generally use the same network, and the network has the same inherent reliability or stability.

The chart that is shown in the following figure can help you determine how to define routing group boundaries.

Determining routing group boundaries

e7a655e7-10d8-440f-8db1-e084bd2dfde1

Defining Routing Group Connectors and Bridgehead Servers

Although all servers communicate with each other directly within a routing group, this is not the case when a server in one routing group needs to communicate with a server in another routing group. To allow servers to communicate with servers in other routing groups, you need to create a routing group connector. Although you can use an X.400 connector or an SMTP connector to connect routing groups, the routing group connector is specifically designed for this purpose and is the preferred method of connecting routing groups in most cases.

By default, all servers within a routing group can send mail over the routing group connector. Servers that are capable of sending mail over a routing group connector are bridgehead servers. A bridgehead server is a combination of an SMTP virtual server and an Exchange server that is responsible for delivering all messages through a connector.

When you create a routing group connector, you have the option of either keeping all the servers as bridgehead servers for that connector, or specifying that only a selected set of servers act as bridgehead servers for that connector. The following table compares the advantages of each approach.

Selecting the number of bridgehead servers in a routing group

Number of bridgehead servers Advantages

All servers in a routing group

  • Provides more efficient message flow because all of the servers in the routing group can directly deliver messages to other routing groups.

  • Capitalizes on configurations where all of the servers in a routing group have the same network connectivity to the servers in other routing groups.

  • Can add complexity in large organizations where all servers communicate in a point-to-point fashion. It can be more difficult to troubleshoot mail flow issues.

  • Direct point-to-point connectivity can provide load balancing.

Only a select few servers in a routing group

  • Makes troubleshooting message flow easier because there are limited points of contact between routing groups.

  • Distributes messaging if you anticipate heavy message flow between routing groups.

  • Allows you to specify server roles of bridgehead servers and mailbox servers in large environments where you do not want mailbox servers handling the traffic sent through a bridgehead server.

  • Makes mail flow more reliable and efficient in those configurations where some servers have better network connectivity than others.

The following figure illustrates the basic components of routing discussed thus far. Figure 5.4 shows message flow between servers within a routing group and between routing groups. It also illustrates a topology that uses only a single bridgehead server in each routing group.

Communication within and between routing groups

9d01e2f8-9211-4f97-853a-7535fc0d4444

When a topology is as simple as that shown in Figure 5.4, you do not have to consider how to best route messages between routing groups. As topologies become more complex, with large numbers of routing groups spread over varying geographical distances, message routing among groups becomes critical.

You configure routing among routing groups by assigning costs (an associated expense for the route based on network availability, network traffic, and administrative requirements) to the routing group connectors that are used by these groups. When a user on a server in one routing group sends mail to a user on a server in another routing group, Exchange uses these costs (part of the link state information that is maintained by Exchange) to determine the most efficient route. Exchange always uses the route with the lowest cost unless a connector or server in that route is unavailable. So that every routing group knows what the various costs are for each connector and the status of those connectors, each routing group has a routing group master that updates and coordinates this information with all of the other servers in a routing group. For more information about routing group masters, see "Designating a Routing Group Master" later in this chapter.

Connecting Routing Groups

When you create a routing group, you designate a group of servers that can communicate directly with one another. For servers in different routing groups to communicate with each other, you need to connect the routing groups.

It is possible to connect routing groups by using either an SMTP connector or an X.400 connector. However, using these types of connectors is generally not recommended. The preferred connection method is a routing group connector because this connector is designed and intended specifically for connecting routing groups.

Note

If you must use an SMTP or X.400 connector between routing groups, do not add an address space on the connector. You should only designate a connected routing group; otherwise, routing will not function correctly.

Routing group connectors are one-way routes for outgoing messages, which means that messages travel outbound to the connected routing group. For two routing groups to communicate, a routing group connector must exist in each routing group to send messages outbound to the other routing group. When you create a connector to a routing group, Exchange displays a message asking if you want to create a routing group connector in the remote routing group so that you can send messages from the remote routing group to the routing group where you are creating the first connector.

Before you create and configure a routing group connector, you should think about the following questions:

  • To which routing group does this connector deliver messages?   This information is critical. Identifying the routing group to which the connector delivers messages establishes the relationship between the sending and receiving routing groups and the rest of your topology. You need to know how the sending and receiving routing groups fit into your topology so that you can determine a cost for the associated connector.

  • What cost should this connector have?   Cost is the variable that Exchange uses to determine the most efficient messaging route. Exchange considers the lowest cost route the most efficient. Exchange uses a more expensive route only if a server or connector is unavailable on the route with the lowest cost. You should assign the lowest costs to the routes with the highest available network bandwidth.

  • Which servers in the routing group can act as bridgehead servers?   Only designated bridgehead servers can send messages across the connector to the connected routing group. The default and preferred setting is to have the servers in the local routing group send mail using this connector. Use this default option when all servers in the routing group can connect directly over the network to the remote bridgehead server and share the same messaging load. Connecting directly to the remote bridgehead server provides more efficient message flow.

    However, you may have better direct network connectivity between specific servers in the local routing group and the designated remote bridgehead server. For example, Server A has a direct connection of 56 kilobits per second (Kbps) to a remote bridgehead server, and Server B and Server C each have a direct connection of 10 megabits per second (Mbps) to the same remote bridgehead server. In this case, you should specify the servers that have the better direct network connectivity (that is, Server B and Server C) as the bridgehead servers, and add those specific servers to a list of allowable bridgehead servers.

    You can configure all servers in the routing group to act as bridgehead servers in one of two ways:

    • Select the default option of Any local server can send over this connector. When you select this option, the connector is always marked as in service or available even if all bridgehead servers become unavailable. This option offers the advantage of generating less link state information because this connector is never marked as unavailable.

    • Select These servers can send mail over this connector and manually add each server in the routing group as a bridgehead server. When you configure your bridgehead servers in this way, if all the bridgehead servers become unavailable, the routing group connector is marked as unavailable. However, using this option can increase the size of your link state table because the fully qualified domain name (FQDN) of each bridgehead virtual server is then written to the link state table. For more information about link state, see Advanced Link State Concepts.

    For more information about evaluating the advantages of using multiple bridgehead servers versus using designated bridgehead servers, see Table 5.3 earlier in this chapter.

  • Should users access public folders that are not available locally using this connector?   By default, public folder referrals are enabled across connectors connecting routing groups. However, network traffic increases when users access a public folder in a remote routing group. If your routing groups are connected by slow network links, or if your network may not be able to handle the additional traffic, disable public folder referrals.

  • What are the remote bridgehead servers to which this connector can send messages?   The remote bridgehead servers are the servers in the connected routing group that receive all messages destined for this routing group. The remote bridgehead servers also receive link state information from the bridgehead servers for the connector.

After considering these questions, you can set your configurations options on the General tab in the Routing Group Connector Properties dialog box. You can address the last question in the above list by specifying remote bridgehead servers on the Remote Bridgehead tab.

For detailed instructions, see How to Configure the Options for a Routing Group Connector and How to Specify a Remote Bridgehead Server for a Routing Group Connector.