220 WAVELENGTH ROUTING OPTICAL NETWORKS
a mixture of packet-switching and circuit-switching. Packet switching is used within the IP networks; circuit-switching is used within the optical network, where a circuit is a
lightpath or subrate channel if traffic grooming is used. The peer model uses the second control plane architecture described above. That is,
the client networks and the optical network are treated as a single network from the point of view of the control plane. The generalized MPLS GMPLS architecture is used in the
control plane. GMPLS is an extension of MPLS for MPLS, see Chapter 6; for GMPLS, see Section 9.5. The IP and the optical networks run the same IP routing protocol – OSPF
with suitable optical extensions. Consequently, all of the optical nodes and IP routers maintain the same topology and link state information. An IP router computes an LSP
end-to-end, which is then established using the label distribution protocols CR-LDP or RSVP-TE see Chapter 7, appropriately extended for GMPLS.
In the overlay model, the optical network uses the first control plane architecture described above see also Figure 9.16. An IP client network is connected to the optical
network via an edge IP router which has an optical interface to its ingress optical node, i.e. the optical node to which it is directly attached. Before an edge IP router can transmit
over the optical network, it has to request a connection from its ingress optical node. This is done by using a signaling protocol defined over a UNI. A connection over the optical
network can be a lightpath permanent or switched or a subchannel. The edge router is not aware of the topology of the optical network; nor is it aware of its control and data
planes. The control plane of the optical network can be based on GMPLS. However, UNI maintains a strict separation of the client networks and the optical network.
Finally, in the augmented model, the IP client networks and the optical network use separate control planes. However, routing information from one network is passed to the
other. For instance, IP addresses from one IP client network can be carried by the optical network to another IP client network to allow reachability. Routing within the IP and
optical networks is separate, but both networks use the same routing protocol. The inter- domain IP routing protocol BGP can be adapted for exchanging information between IP
and optical domains.
9.5 GENERALIZED MPLS GMPLS
The generalized MPLS GMPLS architecture is an extension of MPLS described in Chapter 6. MPLS was designed originally to introduce label-switching paths into the
IP network, and as we saw in Chapter 6, it is also applicable to ATM, frame relay and Ethernet-based networks. The GMPLS architecture was designed with a view to applying
label-switching techniques to time-division multiplexing TDM networks and wavelength routing networks in addition to packet-switching networks.
A TDM network is a network of SONETSDH links interconnected by digital cross connect systems DCS
; see Section 2.5. A DCS terminates the SONETSDH signal on each incoming link, converts it into the electrical domain, and then switches the contents
of some of the virtual tributaries to different outgoing SONETSDH frames. It also drops some virtual tributaries, and adds new ones to the outgoing frames. The outgoing frames
are then transmitted out over the SONETSDH output links of the switch. Aggregation of SONETSDH payloads to a higher SONETSDH level can also be done at the output
links. A circuit-switching connection through such a SONETSDH network can be set up by allocating one or more slots of a SONETSDH frame along the links that make up the
GENERALIZED MPLS GMPLS 221
path see Section 2.5. GMPLS can be used to configure the SONETSDH DCSs, so as to set up a circuit-switching connection.
GMPLS can also be used to set up a lightpath in a wavelength routing optical network. In addition, it can be used to configure an OXC so that to switch the entire optical signal
of an input fiber to an output fiber. In GMPLS, IP routers, ATM switches, frame relay switches, Ethernet switches, DCSs
and OXCs are all treated as a single IP network from the control point of view. There are no UNIs and NNIs, since GMPLS is a peer-to-peer protocol.
GMPLS is an architecture and its implementation requires a signaling protocol. Both RSVP-TE and CR-LDP have been extended to support GMPLS.
In the rest of this section, we describe the basic features of the GMPLS architecture and the extensions proposed to CR-LDP and RSVP-TE.
9.5.1 Basic Features of GMPLS
A GMPLS-capable LSR can support one or more of the following interfaces: 1. Packet-switch capable PSC interfaces: These are the different interfaces used to
receive and transmit packets, such as IP packets, ATM cells, frame relay frames, and Ethernet frames. Forwarding of these packets is based on: an encapsulated label,
the VPIVCI field of the ATM cell header, or the DLCI field of the frame relay frame. 2. Time-division multiplex capable TDM interfaces: They forward data based on the
data’s slots within a frame. This interface is used in a SONETSDH DCS. 3. Lambda switch capable LSC interfaces: They forward data from an incoming wave-
length to an outgoing wavelength. This interface is used in OXCs. 4. Fiber-switch capable FSC interfaces: They forward data from one or more incoming
fibers to one or more outgoing fibers. They are used in an OXC that can operate at the level of one or more fibers.
These four interfaces are hierarchically ordered see Figure 9.17. At the top of the hierarchy is the FSC, followed by the LSC, then TDM, and finally PSC. This order of
the interfaces is used by GMPLS to support hierarchical LSPs. Recall from Section 6.2.4 that MPLS also supports hierarchical LSPs. Consider an LSP that starts and ends at a
packet-switching interface. This LSP can go through several types of networks, where it can be nested together with other LSPs into a higher-order LSP. The high-order LSP can
start and end at a packet-switching interface, a time-division interface, a lambda switch interface, or a fiber-switch interface. In general, the nesting of LSPs into a high-order
LSP is done following the hierarchy of the above four interfaces see Figure 9.17.
An example of a hierarchical LSP is shown in Figure 9.18. Assume that a number of IP routers are connected to a SONETSDH network, which in turn is connected to
Highest level
Lowest level FSC
LSC TDM
PSC
Figure 9.17 The hierarchy of the four types of interfaces.
222 WAVELENGTH ROUTING OPTICAL NETWORKS
IP Router A DCS B IP Router C
OXC A OXC B
1 GbE OC-48
OC-192 32 OC-192
OC-192
Packet LSP1 TDM LSP2
Lambda LSP3 1 GbE
DCS A IP Router B
Figure 9.18 An example of hierarchical LSPs.
a backbone wavelength routing network. The LSP starts at IP router A and ends at IP router C. As can be seen, IP router A is connected to IP router B via a 1-GbE link, and
IP router B is connected to DCS A via an OC-48STM-16 SONETSDH link. DCS A is connected to OXC A via an OC-192STM-64 SONETSDH link. OXCs A and B are
part of a wavelength routing network, and are connected by a single fiber that has 32 wavelengths – with each wavelength carrying an OC-192STM-64 SONETSDH stream.
At the other side of the wavelength routing optical network, OXC B is connected to DCS B via an OC-192STM-64 SONETSDH link, and DCS B is connected to IP router C via
a 1-GbE link.
The interfaces along the path of the LSP from IP router A to IP router C can be easily deduced. The 1-GbE links between IP routers A and B, and DCS B and IP router C have
PSC interfaces. The SONETSDH links between IP router B and DCS A, DCS A and OXC A, and OXC B and DCS B have TDM interfaces. Finally, the link between OXCs
A and B has an LSC interface.
As we move towards the wavelength routing optical network, the capacity of the links increase. This is indicated in Figure 9.18 by using thicker lines. On the other side of
the wavelength routing optical network, the link capacities decrease as we move towards the edge, and this is indicated by decreasing the thickness of the lines. The increase in
the link capacity as we move closer to the backbone network is normal, since the links carry more traffic than those in the edge of the network.
In Figure 9.18, the LSP between IP routers A and C is labeled as packet LSP1. As can be seen, this LSP is nested together with other LSPs in the TDM LSP2, which in turn is
nested in the lambda LSP3. When LSP1 is being established, DCS A will try to allocate bandwidth within its TDM LSP2. If this is not possible, DCS A will establish a new TDM
LSP2 to DCS B. The new TDM LSP2 will be nested within the lightpath lambda LSP3, if bandwidth is available. Otherwise, OXC A will establish a new lightpath to OXC B. If
LSPs 2 and 3 do not exist at the time when IP router A is attempting to establish LSP1, then the establishment of LSP1 will trigger DCS A to establish TDM LSP2, and OXC A
to establish lambda LSP3.
The generalized label request The generalized label request is used to communicate characteristics required to support
the establishment of an LSP. The information required in a generalized label request is shown in Figure 9.19. The following fields have been defined:
GENERALIZED MPLS GMPLS 223
1 2
3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
LSP enc. type G-PID
Switching type Figure 9.19
The information carried in a generalized label request.
• LSP encoding type
: This 8-bit field indicates how the data to be transmitted over the LSP will be encoded. The following values have been defined:
Value Type
1 Packet
2 Ethernet V2DIX
3 ANSI PDH
4 ETSI PDH
5 SDH ITU-T G.707
6 SONET ANSI T1.105
7 Digital wrapper
8 Lambda photonic
9 Fiber
10 Ethernet 802.3
11 Fiber Channel
• Switching type
: An 8-bit field used to indicate the type of switching that should be performed on a particular link. This field is used on links that advertise more than one
type of switching capability. •
Generalized payload identifier G-PID : A 16-bit filed used to identify the payload
carried by an LSP. It is used by the endpoints of the LSP. The following are some of the values specified:
Value Type
Technology Unknown
All 14
Byte synchronous mapping of E1 SONETSDH
17 Bit synchronous mapping of DS1T1
SONETSDH 28
PoS- No scrambling, 16 bit CRC SONETSDH
32 ATM mapping
SONET, SDH 33
Ethernet Lambda, Fiber
34 SDH
Lambda, Fiber 35
SONET Lambda, Fiber
36 Digital wrapper
Lambda, Fiber 37
Lambda Fiber
The generalized label Since the scope of MPLS was widened into the optical and TDM domains, several new
forms of labels are required. The generalized label not only allows for the MPLS-type label that travels in-band with the associated packet, but also allows for labels that identify time
