10.7. Optical Intradomain Routing
Different standards bodies, such as the IETF, the OIF and the ITU-T, are working on enhancing intradomain routing protocols such as OSPF and IS-IS, for routing and topology discovery in networks that consist of optical or other circuit switches. In this section, we discuss the work in progress in different standards bodies, using OSPF as an example.
Over the last several years, the OSPF routing protocol [Moy98] has been widely deployed throughout the Internet. As discussed in the last chapter, OSPF is a link state routing protocol. OSPF specifies a class of messages called LSAs that allow nodes in the network to update each other about their local links. Five different types of LSAs are defined in OSPF, as described in the last chapter.
As discussed earlier in this chapter, routing in IP and optical networks is not quite the same, and hence OSPF needs to be enhanced to route connections in optical networks. For example, in optical networks, the links are always point-to-point. Hence, network LSAs are not required. On the other hand, the link state information contained in router LSAs do not include detailed information on link characteristics, such as resource availability, physical diversity, and so on. Consequently, extensions to LSAs are required. Instead of enhancing the router LSA, these enhancements have been accommodated using opaque LSAs, which provide a generalized mechanism to extend OSPF.
Opaque LSAs consist of a standard LSA header (Figure 10-2) followed by a application-specific information field [Coltun98]. Like other LSAs, the opaque LSA is distributed using the OSPF flooding mechanism. The manner in which an opaque LSA is flooded depends on the flooding scope of the LSA. The following describes different flooding scopes:
Link local flooding: In this case, the LSA is only transmitted over a single point-to-point or broadcast link.
Area local flooding: In this case, the opaque LSA is flooded only in the area where it originated.
Autonomous system (AS) wide flooding: In this case, the LSA is flooded throughout the AS.
Figure 10-2. Link State Advertisement (LSA) Format in OSPF
In the following sections, we describe how opaque LSAs have been used to extend OSPF for optical routing. Specifically, we describe routing within a single OSPF area, as well as routing across multiple areas. As one can possibly anticipate, routing across multiple areas is much more complex than routing within a single OSPF area. Consequently, it does not come as a surprise that the current state of the art for routing in a single area is more mature as compared with routing across multiple areas.
10.7.1. Routing in a Single Area
OSPF enhancements in support of MPLS TE [Katz+02] were already underway when the IETF started its work on routing in optical networks. It turned out that the requirements driven by MPLS TE extensions have a lot of overlap with those driven by routing in optical networks. As a result, these two efforts have been aligned under the umbrella of GMPLS OSPF TE extensions or, simply, GMPLS OSPF-TE. In the following, we discuss this work and how it addresses the requirements for routing in an optical network. A similar initiative is also underway to enhance IS-IS [Callon90, Kompella02b].
The extensions to OSPF have been brought about by extending the notion of a link. In addition to advertising “regular” point-to-point links using router LSAs, OSPF-TE also allows a node to advertise TE links. A TE link is a single or a bundled link with associated characteristics. The properties of the link pertaining to shortest path first (SPF) computation (e.g., link cost) are advertised using router LSAs. The other characteristics of a TE link are advertised using opaque LSAs. Opaque LSAs carrying such TE information are also known as TE LSAs. In summary, a TE link is a “logical” link that has several properties and it may or may not have one-to-one correspondence to a “regular” point-to-point link.
The liveness of a TE link, that is, whether the link is functional, is determined by the liveness of each its component links. A bundled link is said to be alive when at least one of its component links is functional. The liveness of a component link can be determined by using any of several means: routing protocol (e.g., IS-IS or OSPF) Hello[1] messages over the component link, signaling protocol (e.g., RSVP) Hellos (see Chapter 7), link management protocol (LMP) Hellos (see Chapter 6), or from link layer indications (see Chapter 3). Once a bundled link is determined to be alive, it can be advertised using OSPF LSA flooding.
[1] Even if IS-IS/OSPF hellos are run over the all component links, IS-IS/OSPF flooding can be restricted to just one of the component links.
The characteristics of a single component link, include the following [Kompella02a]:
Maximum bandwidth: This parameter specifies the absolute maximum bandwidth usable on the link. This is the capacity of the link.
Unreserved bandwidth: This parameter specifies the amount of bandwidth not yet reserved on the link. Unreserved bandwidth values can be advertised separately for different priority levels. Currently, eight priority levels are supported. For a bundled TE link, the unreserved bandwidth is the sum of unreserved bandwidths of all the component links.
Maximum and minimum connection bandwidth: These parameters determine the maximum and minimum bandwidth that can be allocated to a connection on the link. The maximum bandwidth that can be allocated is less than the unreserved bandwidth on the link. The minimum bandwidth that can be allocated depends on the granularity of switching supported by switching nodes. The maximum connection bandwidth of a bundled link is defined to be the maximum of the maximum connection bandwidth of all the component links.
For example, suppose there are two unreserved OC-192 channels or component links in a TE link. The unreserved bandwidth is then 2 × 10 Gbps or 20 Gbps. The maximum bandwidth that can be allocated to a connection, however, is only 10 Gbps. Assuming that the switches support only STS-48c or 2.5 Gbps switching granularity, the minimum bandwidth that can be allocated to a connection is 2.5 Gbps.
Link protection type: This describes the protection capabilities of the link. Some of the supported protection types are preemptible, unprotected, shared, dedicated 1:1 and dedicated 1+1, as described earlier.
SRLG: This is an unordered list of numbers that are the SRLG identifiers associated with the link.
Interface switching capability descriptor: This includes the following parameters:
Switching capability: This parameter identifies the switching capabilities of the interfaces associated with the link. The following values are relevant to optical switching: Time-Division-Multiplex (TDM) capable, lambda-switch capable, and fiber-switch capable. These were defined earlier.
Switching-capability-specific information: The switching-capability-specific information depends on the switching capability parameter. For example, when the switching capability parameter is TDM, the specific information includes switching granularity, an indication of whether the interface supports standard or arbitrary SONET/SDH concatenation, and so on.
10.7.2. An Example
Consider the network shown in Figure 10-3, consisting of four optical switches, nodes A–D, and six TE links, 1–6. We assume that the switches are TDM-switching capable at STS-48c granularity. Let us also assume that TE links are link bundles with the following properties:
Link 1: 4 OC-48 component links, 2 used and 2 unused, unprotected, and SRLG ID = 123
Link 2: 4 OC-48 component links, 2 used and 2 unused, unprotected, and SRLG ID = 234
Link 3: 4 OC-48 component links, 2 used and 2 unused, unprotected, and SRLG ID = 345
Link 4: 4 OC-192 component links, 2 used and 2 unused, unprotected, and SRLG ID = 456
Link 5: 4 OC-192 component links, 2 used and 2 unused, unprotected, and SRLG ID = 567
Link 6: 4 OC-192 component links, 2 used and 2 unused, unprotected, and SRLG ID = 678
Figure 10-3. Example of an Optical Network Configured as a Single OSPF Area
| LS Type | LS ID | Advertising Node | Comments |
|---|---|---|---|
| Router LSA | Node ID of A | Node A | SPF properties of Links 1 and 5 |
| Router LSA | Node ID of B | Node B | SPF properties of Links 1,2,3, and 4 |
| Router LSA | Node ID of C | Node C | SPF properties of Links 3,4,5, and 6 |
| Router LSA | Node ID of D | Node D | SPF properties of Links 2 and 6 |
Let us assume that the network is configured as a single OSPF area. Consequently, no summary LSAs are required. Because we only have point-to-point links, no network LSAs are necessary. Let us also assume that there are no external routes, and hence no external LSAs are required either. The link state database of each node will consist of only router LSAs and TE LSAs. Because there are four nodes, only four router LSAs, one originated by each node, will be present. There will be twelve TE LSAs, two each for every TE link since both end points advertise the link between them. Table 10-1 shows the router LSAs maintained by each node and Table 10-2 shows the TE LSAs (see also Figure 10-2).
| LS Type | LS ID | Advertising Node | Comments |
|---|---|---|---|
| TE LSA | Link ID of link 1 | Node A | TE properties of link 1 as seen by A |
| TE LSA | Link ID of link 1 | Node B | TE properties of link 1 as seen by B |
| TE LSA | Link ID of link 2 | Node B | TE properties of link 2 as seen by B |
| TE LSA | Link ID of link 2 | Node D | TE properties of link 2 as seen by D |
| TE LSA | Link ID of link 3 | Node B | TE properties of link 3 as seen by B |
| TE LSA | Link ID of link 3 | Node C | TE properties of link 3 as seen by C |
| TE LSA | Link ID of link 4 | Node B | TE properties of link 4 as seen by B |
| TE LSA | Link ID of link 4 | Node C | TE properties of link 4 as seen by C |
| TE LSA | Link ID of link 5 | Node A | TE properties of link 5 as seen by A |
| TE LSA | Link ID of link 5 | Node C | TE properties of link 5 as seen by A |
| TE LSA | Link ID of link 6 | Node C | TE properties of link 6 as seen by C |
| TE LSA | Link ID of link 6 | Node D | TE properties of link 6 as seen by D |
Now let us examine the TE LSA pertaining to link 1, advertised by node A. Among other items, it will include the following pieces of information:
Maximum bandwidth (i.e., absolute capacity) is 4 × 2.5 Gbps or 10 Gbps since there are four OC-48 component links in the TE link.
Maximum available or unreserved bandwidth is 2 × 2.5 Gbps or 5 Gbps since only two OC-48 component links are available.
Maximum connection bandwidth is 2.5 Gbps since the component links are OC-48.
Minimum connection bandwidth is 2.5 Gbps since the switching is at the granularity of STS-48c.
The protection type of the links are “unprotected” and the associated SRLG list contains only one SRLG ID, that is, 123.
The switching capability of the interface is TDM since the switches are TDM-capable. Switching granularity is STS-48c.
Using the information contained in the router and the TE LSAs, each node can easily construct a topology database that contains the information shown in Table 10-3. Note that Table 10-3 shows only a subset of different fields; the actual topology database will contain other information fields. As shown in Table 10-3, the topology database contains detailed information about different TE links between nodes. This information is used by the route computation module to compute explicit routes to different destinations.
The link information presented in Table 10-3 is sufficient to compute unprotected and 1 + 1 protected paths. It is not, however, sufficient to compute shared mesh protected paths (Chapter 8). Recall that under shared mesh protection, multiple backup paths may share protection links as long as the corresponding primary paths do not have any common SRLGs. Note that Table 10-3 does not provide any information about shared backup links and the connections that can potentially use them.
In the absence of any additional information, simple heuristics can be used to compute shared backup paths. For example, a physically diverse protection path can be computed first. Then, sharing can be attempted on as many links as possible during signaling, using information available at each node along the protection path (see [Sengupta+02]). We do not consider this topic further here.
10.7.3. Routing in All-Optical Networks
Routing in an all-optical network without wavelength conversion raises several additional issues that have not been addressed in the OSPF enhancements discussed earlier. First, since the route selected must have the chosen wavelength available on all links, this information needs to be considered in the routing process. Consequently, wavelength availability information needs to be advertised by the routing protocols. An alternative is to “hunt” for available wavelengths during signaling using the GMPLS Label Set feature (see Chapter 7), but relying on this alone may not be efficient.
| Node A | Node B | Node C | Node D | |
|---|---|---|---|---|
| Node A | Link 1: Max capacity 10 Gbps; available capacity 5 Gbps; switching capability SONET, switching granularity 2.5 Gbps; unprotected; SRLGs {123} ; max connection bandwidths 2.5.Gbps; min connection bandwidth 2.5Gbps. | Link 5: Max capacity20 Gbps; available capacity 10 Gbps;switching capability SONET; switching granularity 2.5 Gbps; unprotected; SRLGs {567} ; max connection bandwidth 10 Gbps; min connection bandwidth 2.5Gbps. | ||
| Node B | Link 1: | Link 3: Max capacity 10 Gbps; available capacity 5 Gbps; switching capability SONET; switching granularity 2.5 Gbps; unprotected; SRLGs {345} ; max connection bandwidth 2.5.Gbps; min connection bandwidth 2.5.Gbps. | Link 2: Max capacity 10 Gbps; available capacity 5 Gbps; switching capability SONET; switching granularity 2.5 Gbps; unprotected; SRLGs {234} ; max connection bandwidth 2.5 Gbps; min connection bandwidth 2.5Gbps. | |
Link 4: Max capacity 20 Gbps; available capacity 10 Gbps; switching capability SONET; switching granularity 2.5 Gbps; unprotected; SRLGs {456} ; max connection bandwidth 10 Gbps; min connection bandwidth 2.5Gbps. | ||||
| Node C | Link 5: | Link 3: | Link 6: | |
| Link 4: | ||||
| Node D | Link 2: | Link 6: Max capacity 20 Gbps; available capacity 10 Gbps; switching capability SONET; switching granularity 2.5 Gbps; unprotected; SRLGs {678} ; max connection bandwidth 10 Gbps; min connection bandwidth 2.5Gbps. |
Next, link-specific information for each type of impairment that has the potential of being limiting for some routes needs to be advertised. Routing constrains imposed by PMD and ASE fall in this category. Other link-dependent information includes span-length, cross talk, and so forth. In addition to the link-specific information, bounds on each of the impairments need to be quantified. Since these are system dependent, they are determined by the system designer's impairment allocations.
Finally, as optical transport technology evolves, the set of constraints that need to be considered may change. The routing and control plane design should therefore be as open as possible, allowing parameters to be included as necessary.