Connection network concepts

To understand connection networks, consider the following concepts:

A connection network provides direct communication between all nodes on a shared access transport facility (such as an IP network) with the addition of just one definition on each node; namely, a link to the named connection network.
A connection network is also referred to as a virtual node (VN) or a virtual routing node (VRN).
Each EE connection network represents a single IP network of a given address family (IPv4 or IPv6) that is used for communicating between a set of EE-enabled partners.
Connection network names are specified as fully qualified names of the form netid.vnname. All nodes on the same shared media that also have connection network functions can specify the same netid.vnname, which enables the node computing the route to recognize that all of these nodes are on the same shared media and can bringing up dynamic links for LU-LU sessions.
Connection network links are not used for CP-CP sessions. End nodes need to predefine a link to their network node server (and usually to their backup network node server) and can also predefine links to other nodes. The predefined link to the network node server is used for CP-CP sessions. Dynamically created connection network links are used only for LU-LU sessions and are typically brought down when the sessions end. (One exception to this is when a DYNTYPE=VN model is coded to change the setting of the DISCNT parameter for connection network PUs.) Similarly, network nodes also predefine links to adjacent network node partners for CP-CP sessions, with connection network links used only for LU-LU sessions.
All participants on a local connection network (VNTYPE=LOCAL) must be located within the same APPN topology subnetwork. In a global connection network (VNTYPE=GLOBAL) nodes located in different APPN topology subnetworks can share the same connection network, thereby providing the capability for direct communication between the nodes defining the global VRN without requiring that all session traffic traverse the extended border nodes (EBNs) that interconnect those subnetworks. The EBNs must still be defined and participate in the session setup process, however.