Virtual Concatenation DATA OVER SONETSDH DOS

48 ATM NETWORKS

3.1 INTRODUCTION

The ATM architecture was designed with a view to transmitting voice, video, and data on the same network. These different types of traffic have different tolerance levels for packet loss and end-to-end delay, as shown in Table 3.1. For instance, packets containing voice have to be delivered on time so that the play-out process at the destination does not run out of data. On the other hand, the loss of some data might not necessarily deteriorate the quality of the voice delivered at the destination. At the other extreme, when transporting a data file, loss of data cannot be tolerated since this will compromise the file’s integrity, but there is no stringent requirement that the file should be delivered as fast as possible. The ATM architecture is based on the packet-switching principle and is connection- oriented. That is, in order for a sender to transmit data to a receiver, a connection has to be established first. The connection is established during the call setup phase, and when the transfer of data is completed, it is torn down. There is neither error control nor flow control between two adjacent ATM nodes. Error control is not necessary, since the links in a network have a very low bit-error rate. In view of this, the payload of a packet is not protected against transmission errors. However, the header is protected in order to guard against forwarding a packet to the wrong destination The recovery of a lost packet or a packet that is delivered to its destination with erroneous payload is left to the higher protocol layers. The lack of flow control requires congestion control schemes that permit the ATM network operator to carry as much traffic as possible without losing too many cells.

3.2 THE STRUCTURE OF THE HEADER OF THE ATM CELL

The ATM packet is known as cell, and it has a fixed size of 53 bytes. It consists of a pay- load of 48 bytes and a header of 5 bytes see Figure 3.1. Several considerations – mostly related to the efficient transport of voice – led ITU-T to decide on such a small fixed- size packet. Two different formats for the cell header were adopted, one for the user network interface UNI and a slightly different one for the network-network interface NNI. The Table 3.1 Tolerance levels by traffic type. Traffic Type Tolerance Levels Packet-loss sensitive Delay sensitive Voice Low High Video Moderate High Data High Low Header Payload 48 bytes 5 bytes Figure 3.1 The ATM cell. THE STRUCTURE OF THE HEADER OF THE ATM CELL 49 1 2 3 4 5 6 7 8 GFC VPI VPI VCI VCI VCI PTI CLP HEC Information payload 1 2 3 4 5 6 . . . 53 b y t e s UNI cell format 1 2 3 4 5 6 7 8 VPI VPI VCI VCI VCI PTI CLP HEC Information payload 1 2 3 4 5 6 . . . 53 NNI cell format b y t e s Figure 3.2 The structure of the cell header. UNI is concerned with the interface between an ATM end device and the ATM switch to which it is attached. An ATM end device is any device that can be attached directly to an ATM network and that can transmit and receive ATM cells. The NNI is used between two ATM switches belonging to the same network or to two different networks. The format of the cell header for these two interfaces is shown in Figure 3.2. As we can see, these two formats differ only in the first field. We now proceed to discuss in detail each field in the header. Understanding the meaning of these fields helps to better understand the ATM network architecture. The generic flow control GFC field permits multiplexing of transmissions from several terminals on the same user interface. It is used to control the traffic flow from the end device to the network. An ATM connection is identified by the combined virtual path identifier VPI and virtual channel identifier VCI . Such a connection is referred to as a virtual channel connection VCC . The VPIVCI field is 24 bits in the UNI interface and 28 bits in the NNI interface. The VPI field is 8 bits in the UNI interface and 12 bits in the NNI interface. Therefore, in a UNI interface, there can be a maximum of 256 virtual paths, and in an NNI interface, there can be a maximum of 4096 virtual paths. In each interface, there can be a maximum of 65,536 VCIs. A VPI can be assigned to any value from 0 to 255. VCI values are assigned as follows: 0 to 15 are reserved by ITU-T, 16 to 31 are reserved by the ATM Forum, and 32 to 65,535 are used for user VCCs. The combined VPI and VCI allocated to a connection is known as the connection identifier CI. That is, CI = {VPI, VCI}. A virtual channel connection between two users consists of a path through a number of different ATM switches. For each point-to-point link that lies on this path, the connection is identified by a different VPIVCI. That is, each VPIVCI has local significance and is translated to a different VPIVCI at each switch that the cell traverses. This operation is referred to as label swapping since the connection identifier is also known as a label, a term adapted later on in MPLS. Label swapping involves a look-up in the switching