background image
Tech Note #22: Asynchronous Transfer Mode (ATM)  Addresses
1998 Bionic Buffalo Corporation; All Rights Reserved.
       14 September 1998
1 of 12
Bionic Buffalo Tech Note #22:
Asynchronous Transfer Mode (ATM) Addresses
last revised Monday 14 September 1998
©1998 Bionic Buffalo Corporation. All rights reserved.
Tatanka and TOAD are trademarks of Bionic Buffalo Corporation.
Asynchronous Transfer Mode (ATM) cells each contain addresses, used to route the cells
through the network. This paper introduces the two parts of the address (VPI and CPI), and
provides a general description of how they are used in the context of the overall architecture.
Local Addresses: VPI & CPI
There are two kinds of interfaces between nodes in an ATM network: the User-Network
Interface (UNI) and the Network-Network Interface (NNI). (Each of these has variations, so not
every interface of one type is compatible with every interface of the same type.) These are two
kinds of software/protocol interfaces; hardware is a separate issue.
NNI is used between switches within networks, usually large-scale public or private networks.
Some switches use UNI, rather than NNI. Other switches use UNI for connections to the
outside of the network, and NNI for connections to other switches within the network.
UNI is used everywhere else. Most ATM equipment uses some form of UNI, including
workstations, telephones, faxes, PBXs, and other devices not part of the backbone or major
switching systems. Almost all customer equipment uses UNI, including use for connections to
the phone company or other external carrier.
ATM messages are encapsulated into packets, called cells. Each cell contains a virtual channel
 (VCI) and a virtual path identifier (VPI), which is used to route the cell (if outgoing)
or identify the cell’s source (if incoming). At the NNI, 12 bits are available for the VPI. At the
UNI, 8 bits are available for the VPI. For either NNI or UNI, 16 bits are available for the VCI.
Although the packet format allows VPIs and VCIs of these sizes, most equipment does not
allow all bits to be used, and sets the unused (high-order) bits to zero.
A virtual path, identified by the VPI, is a route between two nodes in the network. A path may
pass through an intermediate node, without terminating. (As an example, see the path from A1
to C1 in the illustration, below.) Each virtual path may contain one or more virtual channels,
each identified by a VCI. (For example, the path from A1 to C1 contains virtual channels U and
V.) Virtual channels are one-way. Two-way communication requires a pair of virtual channels.