10.2. Differences between IP and Optical Routing
Although IP routing protocols are being extended for routing connections in optical networks, there are major differences between routing in circuit switched optical networks and routing in packet switched IP networks.
IP routing involves both control and data plane functionality. The function of the control plane is twofold: to distribute topology information throughout the network (link state routing) and to compute a forwarding table from the topology information. The actual forwarding of IP packets, using the forwarding table, is the function of the data plane. No connection is established ahead of time, and packets are forwarded hop by hop from the source to the destination. In optical networks, as in other circuit-switched networks, the data plane is not involved in connection routing. In these networks, end-to-end connections are explicitly established based on the network topology and resource information. Once a connection is established, data is transferred over that connection without further involvement of the routing engine.
In IP networks, the routing protocols are intimately involved with data plane forwarding decisions and hence their failure adversely impacts the services offered to end users. Due to the separation of control and data planes, routing protocol failures in optical networks do not adversely impact existing connections. This does not imply routing is less critical in the optical case, only that its service-impacting effect is secondary. For example, topology and resource status inaccuracies will affect the establishment of new connections, but will not (and should not) cause an existing connection to be torn down.
Since a connection has to be established and appropriate resources have to be reserved in advance of data transfer, routing in an optical network requires knowledge of the availability of different resources in the network. Current versions of intradomain IP routing protocols do not handle resource availability information. Recent routing extensions for IP traffic engineering address this issue. Further enhancements are necessary to handle detailed resource availability information required for routing connections in optical networks. Also in optical networks, a connection establishment could be blocked if resources are not available, whereas some level of overloading (and temporary congestion) is tolerated in IP networks. In other words, statistical multiplexing is a feature not applicable in optical networks.
Another difference between routing in IP and optical networks is that IP routing is hop-by-hop while routing in optical network is typically source-directed. In other words, with IP routing each node on the path from a source to a destination independently decides the next hop to forward a packet. Hence, it is important that all nodes in the network have consistent topological view and they all use the same route computation algorithm. In optical networks, the source node is responsible for computing the entire path from the source to the destination. Hence, as long as the source has the right topological information of the network, it can compute the best path. Also, it is not necessary for all nodes to use the same route computation algorithm.
Finally, another important difference in routing arises due to protection and restoration requirements. In IP networks, traffic is typically forwarded over the shortest path. If there is a failure in the network, the routing scheme discovers alternate paths and packets are routed around the failure. In other words, network failures are handled in a reactionary fashion. On the other hand, one of the basic features of optical networks is the protection of connections using precomputed and often preprovisioned, facility-diverse backup paths (see Chapter 8). Facility diversity ensures that the working and the protection paths are not concurrently affected by the same failure. In order to compute facility-diverse working and protection paths, switches in the network need to have access to physical plant information. The physical diversity information is not available in IP routing.
In the following, we discuss some of the network characteristics that make routing in optical networks different from routing in IP networks. We also address how these specific characteristics can be abstracted and distributed using the mechanisms provided by link state IP routing protocols.