7.1. Introduction
Connection provisioning is the act of establishing a connection between two end points in an optical network. Provisioning has traditionally been in response to a service request received by the optical network operator from an external user. Recent developments in automating this process allow such requests to originate from user equipment directly. In either case, connection provisioning requires the identification of the two connection end points and the desired connection attributes. As an example, Figure 7-1 illustrates an optical network and some of the parameters specified during connection provisioning initiated using a traditional management system. In this figure, all the ports are assumed to be OC-48, and the A-end and Z-end denote the connection end points (which are indicated by <node, port> pairs). After it is initiated, the provisioning process consists of two major steps: First, the specified parameters are used to compute a route for the connection within the optical network. The connection route is a function of both the connection parameters and the resources available in the network. The route computation process results in the identification of the sequence of nodes and links in the connection path. Next, the connection is established along the computed route by activating the cross-connects in the appropriate network elements (NE). This is the part of interest in this chapter.
Figure 7-1. An Optical Network and Connection Parameters
Connection establishment can be realized broadly in one of two ways. Under the first method, each NE in the connection path may be instructed separately by the management system to establish the appropriate cross-connect. In this case, the management system is responsible for computing the complete connection route. This method is illustrated in Figure 7-2 for the connection-provisioning example shown in Figure 7-1. The communication between the management system and the NEs may use protocols such as TL-1 or SNMP, as described in Chapter 13. The second method is to use control communication between NEs in the connection path. This communication is referred to as signaling, whereby control messages flow from the source NE to the destination NE via intermediate NEs. A signaling protocol defines the syntax and semantics of control messages, as well as rules that govern the transmission, reception, and processing of such messages at each NE. Under many signaling protocols, provisioning requires message flow in two or more passes back and forth between the source and the destination NEs. Figure 7-3 illustrates signaling message flow for provisioning the same connection shown in the previous figures. Even with signaling, connection provisioning may have to be initiated from a management system. The management system in this case triggers the source NE to set up the connection. This NE may compute the connection route, or the route may be computed by the management system and supplied with the provisioning command.
Figure 7-2. Computed Route and Centralized Provisioning
Is one method better than the other? Equipment vendors who use the first method swear by the efficiency and reliability of the method. They point out that a highly reliable management system is required in all networks and that the system must have connectivity to all NEs regardless of whether signaling is used for connection establishment. Furthermore, such a system maintains complete network state information and thus using it for provisioning seems quite natural. On the other hand, the proponents of signaling point out that the management systems are already being overloaded with various tasks (e.g., performance monitoring), and additional network control tasks would strain the system and limit its scalability. Furthermore, signaling capability allows the NEs to speedily react to certain types of failures that require connection reprovisioning (e.g., node failures). Thus, both types of provisioning are here to stay. The focus of this chapter, however, is on provisioning using signaling. Chapter 13 covers management systems.
In this regard, our descriptions so far have been somewhat simplistic in order to explain the basic functionality. In reality, the provisioning problem could be more complex. Specifically, the following issues may arise:
End-to-end connection provisioning in a multivendor composite network: The concept of control domains was described in Chapter 5. In practice, a large optical network may consist of multiple control domains, each consisting of equipment manufactured by a different vendor. Indeed, the provisioning capability within each control domain could be different (e.g., management system based provisioning in one control domain and signaled provisioning in another). A connection in such a composite network consists of connection segments within and between control domains as described in Chapter 5. Provisioning such a connection requires the integration of the control domain provisioning methods. This is indeed the goal of standardized signaling procedures. Different architectural models developed for this purpose are described in the next section.
Connection modification: Parameters of a provisioned connection may be modified by the user of the connection. Effecting the modifications within the network without affecting the existing traffic flow on the connection is called nondisruptive connection modification. This may require special procedures.
Networks with hierarchical routing: A large optical network may employ distributed hierarchical routing. Connection provisioning in such a network is similar to provisioning connections in the presence of subnetworks. Specifically, an end-to-end connection has to be provisioned as a series of connection segments, each spanning a region of the network. This is illustrated under the PNNI signaling procedures described in section 7.6.
Connection hierarchy: The concept of layer networks was described in Chapter 5. Specifically, a “link” between two NEs in one (client) layer network could actually be a connection in an underlying (server) layer network. Provisioning a connection in the former network could thus involve provisioning a connection in the latter network. For instance, consider two metro optical networks attached to a core optical network. Suppose the metro networks support STS-1 connection granularity while the core network provides line-transparent STS-48-level switching granularity. Provisioning an STS-1 connection between a node in one metro network and a node in another network might require the provisioning of an STS-48 line-layer connection between two border nodes in these networks, and then accommodating the STS-1 connection within the larger connection. When signaling is used for provisioning, the larger connection may be provisioned dynamically.
Provisioning protected connections: When a connection is protected, a primary and a protection path must be provisioned at the same time. Signaling procedures for provisioning protected connections are described in the next chapter.
Signaling procedures related to provisioning have been standardized by the ITU-T, the OIF and the IETF. Specific architectural models have been developed by these bodies. ITU-T recommendation G.7713 describes distributed call and connection management models pertaining to optical networks. While somewhat abstract, this recommendation formalizes a number of concepts related to signaling. These are described in the next section. The Generalized Multi-Protocol Label Switching (GMPLS) architecture is the product of the IETF, dealing with signaling-related concepts pertaining to optical networks. This is described in section 7.3.
Architectural models are only half the story when it comes to signaling. The actual details are found in signaling protocols themselves. Two protocols of interest are the GMPLS RSVP-TE and the PNNI protocols. The former protocol was originally designed for MPLS networks, but has been adapted to fit both the G.7713 and the GMPLS architectural models. The OIF has also used it as one option for User-Network Interface (UNI) signaling. The latter was designed originally for signaling in ATM networks, but it has been successfully deployed in operational optical networks. It has also been adapted to support the G.7713 architectural framework. These protocols are described in sections 7.5 and 7.6.