![[previous]](../../../graphics/l_arrow.gif){kind=icon}
![[next]](../../../graphics/r_arrow.gif){kind=icon}
APPN ISR routing works differently from other protocol routing architectures. Unlike other protocols such as IP, you do not configure static routes using APPN. Instead, network nodes maintain a directory of LU resources (and more importantly, the location of the LU resources) available in their domains. When an originating LU requests a session to a destination LU, the location of that destination LU is discovered by checking the directories on the network nodes. The actual route is determined using the APPN class of service tables. For more information on how APPN class of service tables determine the best route to take in an APPN network, see the Configuring APPN Class of Service chapter.
This section describes the following major topics regarding APPN concepts and how APPN network nodes facilitate APPN routing:
This section describes briefly the different node types defined on an APPN network. For more detailed conceptual information, see the IBM document, APPN Architecture and Product Implementations Tutorial and the other documents listed in "IBM APPN References" later in this chapter.
Nodes in an APPN network are divided into the following three types:
Figure 161 is a sample of a network topology with different devices acting as different types of nodes.
Figure 161 Node Types on an APPN Network
Network nodes provide routing services and directory services for LUs on network nodes and end nodes. When a session request is initiated by a LU in the network node, an end node or LEN end node, the network node tries to locate the destination LU either on its own nodes or by querying other nodes. After the LU is located, the network node determines the best route to the destination LU according to the class of service for that session.
When a network node is added to an APPN network, the node learns network topology information from active adjacent network nodes. A network node exchanges network topology information with adjacent network nodes only when there is a change to the network topology.
When used in APPN networks, 3Com bridge/routers can function only as a network node, and does not support local LUs for application programs. Other devices that can serve as network nodes in an APPN environment include the following IBM platforms:
This is not a complete list; other products may also be able to serve as network nodes.
End nodes provide limited directory and routing services for their local LUs. End nodes establish Control Point-to-Control Point (CP-CP) sessions with an adjacent network node so that LUs on the end node are available on the APPN network. The end node can also establish sessions to other LUs in the network.
The end node selects a network node to serve as its network node server and registers its LUs with the network node. By registering the end node's local resources with the network node, the network node can route any session requests from a remote node to the end node's LU. End nodes can have active connections to more than one network node at the same time, but only one network node can serve as the end node's network node server at one time.
Devices that can act as end nodes in an APPN environment include the following IBM platforms:
Low-entry networking (LEN) end nodes are different from normal end nodes in that they cannot establish CP-CP sessions with a network node. As a result, LEN end nodes cannot register their resources with the network node; these resources must be predefined on the network node.
If the LEN end node has only one LU, then that LU is learned dynamically by the network node. However, if the LEN node has more than one LU, all LUs in addition to the first one must be statically defined in the network node's directory. For more information on defining LEN end node resources, see "Adding Entries" earlier in this chapter.
Many devices that are normally network nodes or end nodes can also be LEN end nodes, depending on how they are configured. Examples of devices that can be LEN end nodes in APPN networks include the following:
Non-IBM personal computers can also serve as LEN end nodes.
The primary difference between network nodes and end nodes is how each node type operates. Table 37 compares the basic differences between node types (note that LEN end nodes are a specific type of end node). For more detailed information regarding node type functionality, see the IBM document, APPN Architecture and Product Implementations Tutorial, in "IBM APPN References" later in this chapter.
| 1
End node maintains a directory of its own LUs only.
|
The role of the network node is to provide the network services for the end nodes in its domain. It also provides the directory database that lists LUs in the local network node domain, so that the LUs can be discovered by other network nodes in the network.
The network node maintains a directory database of information for two types of nodes:
Figure 162 is the conceptual difference between the local topology and network topologies known by the network node.
Figure 162 Local Topology and Network Topology for the Network Node
The network node acts as the network node server for the end nodes in its domain. The network node server provides the following services to end nodes:
These services locate network resources in the APPN network, and pass the information onto the end node.
These services calculate the best route between the origin and destination LUs based on the required class of service. For more information on how class of service calculates routes, see the Configuring APPN Class of Service chapter.
An end node can have links to more than one network node. However, only one network node can act as the end node's network node server at one time.
The network node maintains two databases:
These databases are LU resources on end nodes in the network node's local domain. These databases can be LUs on end nodes that were learned dynamically by the network node, or LUs on LEN end nodes that were statically defined in the directory.
This database maintains information regarding all network nodes and the TGs between them. The network nodes and associated TGs together make up the APPN network backbone.
APPN is a point-to-point protocol, which means that links are established between two single partner nodes. The end node maintains a direct link to the network node server.
After you configure a link from the network node to an adjacent end node, the following processes can take place:
1 . The end node calls on the network node to set up a CP-CP session (this does not apply to LEN end nodes).
2 . The end node "registers" its LUs with the network node by sending information from the end node's local directory database to the network node.
3 . After the network node receives the directory information, the directory entries are stored in the network node's local directory database.
These entries are temporary entries in the network node's directory database, and will change depending on how resources change on the end node. For example, if a resource on the end node is added or deleted, the information is sent to the network node's local directory database to be updated.
As long as the end node maintains a CP-CP session with the network node, the end node's resources will be registered in the network node's local database directory. After a CP-CP session is deactivated between the end node and the network node, the end node's registered entries in the network node's directory database are automatically deleted.
Figure 163 illustrates how this process works.
Figure 163 End Node Resource Registration into Network Node's Directory
When end nodes initiate a request for a session with a destination LU, the end node requests that the network node allocate a session to the destination LU. The network node then consults the local directory database to discover if the LU is in its domain. If the destination LU is within the network node server's domain, the network node can send the session request directly to the destination LU. However, if the destination LU is on an end node or network node, the local node may send a Locate request first. When the local node receives a positive response to the locate request, it forwards the BIND request.
Figure 164 is an example of discovering the destination in the local topology. In this example, LU1 on end node "Red" wants to initiate a session with LU6. This example assumes that CP-CP sessions are up between end node "Red" and end node "Green." Based on the examples shown in the figure, the following steps take place:
1 . LU1, the originating LU on end node "Red," initiates a Locate request to network node A, requesting the location of LU6, the destination LU.
2 . The network node checks the local directory database for the location of LU6.
3 . The directory database discovers a directory entry for LU6, which shows it is located on end node "Green," and that this end node is within the network node's local domain.
4 . The network node sends a Locate request to "Green," verifying that LU6 is still available.
5 . "Green" sends a "locate positive" response to the network node.
6 . The network node forwards the response to "Red."
7 . LU1 sends a BIND to LU6 to begin process for a logical unit-to-logical unit (LU-LU) session.
Figure 164 
Discovering a Destination LU in the Local Directory
If the destination LU is not within the network node's local domain, the network node then sends locate requests to adjacent network nodes. These network nodes in turn check their directories to see if the destination LU is in their local domains.
Figure 165 is an example of discovering the destination LU in the larger network topology.
Figure 165 
Discovering a Cross-Domain LU
In this example, since the destination LU is not within the local domain, the network node server sends Locate requests to adjacent network nodes. Those network nodes check their local databases for the destination LU, and if they do not find it, they in turn forward the Locate request to other adjacent network nodes. This process continues until the network node server for the destination LU finds the directory entry in its local directory database, and then forwards the information back to the first network node. The directory entry for the destination LU is then cached in the first network node's directory, in case that LU needs to be located again.
The process of locating LUs using directory services differs from the process of calculating the actual routes to those LUs. Routing is handled by topology and routing services, and routes are determined through the use of class of service tables. For more information on how class of service works, see the Configuring APPN Class of Service chapter.