technologies paving the way to packet-based networks. In contrast to old analog networks, packet-based digital networks can be either connectionless or connection oriented, can have a control plane for the automation of some functions, can implement various resilience schemes, can perform a number of network services supporting users' applications, and so on. The essential ideas are explained in this section as a background for the entire chapter. Digital networks can transfer information

more »

... n nodes by means of two fundamental paradigms: circuit switching or packet switching. . In circuit-switched networks, data are organized in continuous, uninterrupted bit streams. In this mode of operation, a dedicated physical link between a couple of nodes is established. Before starting the data transfer on a specific connection, the connection itself must be "provisioned"; that is, the network switching nodes must be configured to provide the required physical link. This implies an exclusive allocation of network resources for the whole duration of the connection. Such a task is usually performed by dedicated elements belonging to the network control system; network resources are released when the connection ends. This is the way that the plain old telephony service (POTS) has been working so far. The private reservation of network resources prevents other connections from using them while the first one is working, and may lead to inefficient network use. . In packet-switched networks, data are organized in packets of finite length that are processed one by one in network nodes and forwarded based on the packet header information. In this network scenario, each packet exploits switching and transmission devices just for the time of its duration, and these network resources are shared by all packets. This process of packet forwarding and aggregation is called statistical multiplexing and represents the major benefit of packet-switched networks with respect to the circuit-switched networks in terms of network exploitation efficiency. Typical examples of circuit-switching and packet-switching technologies are synchronous digital hierarchy (SDH) and Ethernet respectively. Packet-switched networks can, in turn, work in connectionless or connection-oriented network modes. . In the connectionless network mode, packets are forwarded hop by hop from source node to destination node according to packet header information only, and no transfer negotiation is performed in advance between the network nodes involved in the connection; that is, the source node, optionally the intermediate node(s) and the destination node. . In the connection-oriented network mode, packet transfer from source node to destination node is performed through defined resource negotiation and reservation schemes between the network nodes; that is, it is preceded by a connection set-up phase and a connection usage phase, followed by a connection tear-down phase. Typical examples of packet-switched connectionless and connection-oriented network protocols are Internet protocol (IP) and asynchronous transfer mode (ATM) respectively. 2 The Emerging Core and Metropolitan Networks The main characteristic of the connectionless network mode is that packets are routed throughout the network solely on the base of the forwarding algorithms working in each node; hence, packet routes may vary due to the network status. For instance, cable faults or traffic overloads are possible causes of traffic reroute: in the connectionless network mode, the new route of a packet connection is not planned in advance and, in general, is unpredictable. On the contrary, in the connection-oriented network mode, the route of any connection is planned in advance and, in the case of faults, traffic is rerouted on a new path that can be determined in advance. Since route and rerouting have strong impacts on the quality of a packet connection, the two network modes are used for different network services depending on the required quality and the related cost. Network Layering The functions of a telecommunication network have become increasingly complex. They include information transfer, traffic integrity and survivability aspects, and network management and performance monitoring, just to mention the main ones. To keep this growing complexity under control and to maintain a clear vision of the network structure, layered network models have been developed. According to these models, network functions are subdivided into a hierarchical structure of layers. Each layer encompasses a set of homogeneous network functions duly organized for providing defined services to the upper layer, while using the services provided by the lower layer. For example, in an Ethernet network, the physical layer provides data transmission services to the data link layer. To define transport network architectures, it is essential to start from the description of the lowest three layers [1]: network, data link, and physical layers: It is commonly agreed that the Open System Interconnection (OSI) model is an excellent place to begin the study of network architecture. Nevertheless, the network technologies commercially available do not map exactly with the levels described in the OSI basic model. General Characteristics of Transport Networks Data Plane, Control Plane, Management Plane The layered network models encompass all network functions related to data transfer. However, modern transport networks are often provided with additional functions devoted to network management and automatic network control. Hence, the totality of network functions can be classified into three groups named planes: the data plane, the management plane and the control plane. The functions that characterize each plane are summarized below. . Data plane. The data plane aims at framing and carrying out the physical transportation of data blocks to the final destination. This operation includes all transmission and switching functions. . Control plane. The control plane performs the basic functions of signaling, routing and resource discovery. These are essential operations to introduce automation on high level network functions such as: connection establishment (i.e., path computation, resource availability verification and connection signaling set-up and tear-down), reconfiguration of signaled connections and connection restoration in case of network faults. . Management plane. The management plane performs management functions like alarm reporting, systems configuration and connection provisioning for data and control planes. The complexity of the management plane depends strongly on the availability of a control plane. For example, the management plane of traditional circuit-switched public switched telephone networks is more cumbersome than transport networks with a control plane, since, in the latter case, certain tasks (e.g., connection provisioning and restoration) are carried out by the control plane itself.