Key to the seamless integration of GMPLS--IP and WDM)

The key to seamless combination of GMPLS--IP and WDM
In recent years, with the rapid development of Internet technology, the traffic of data services represented by IP has rapidly increased and has gradually become the mainstream of network services. However, the existing transmission network is voice optimized. to enable it to carry data services efficiently, new technologies are required.
To transmit data services, the existing transmission network adopts a layer-4 structure, as shown in 1a): IP over ATM over SDH over WDM. The IP layer is used to carry services. The ATM layer is used to integrate multiple services and provide quality assurance for each service. The SDH layer is used for fine-grained bandwidth allocation, it also provides a reliable protection mechanism for business transmission. The WDM layer is used to provide large-capacity transmission bandwidth. Although this layer-4 structure transmission method can ensure data service transmission, there are many problems in use.
First, the layer-4 structure has a "bottleneck" effect. In this structure, bandwidth allocation is very troublesome. Not only does it take a long time for manual configuration, But bandwidth is limited by the available bandwidth of each layer of equipment. Even if the vast majority of devices have available idle bandwidth, the bandwidth bottleneck of any device on any layer may limit the bandwidth or capacity expansion of the entire network. At the same time, failure of any layer of equipment will affect the stability of the entire network, followed by low transmission efficiency. Because both ATM and SDH have a large amount of frame head overhead, it directly affects the transmission efficiency of data services. For example, for a 10 Gb/s STM-16 with a net load of 9.6 Gb/s), when a layer-4 structure is used to carry the IP service, about 2.4 Gbit/s of bandwidth is used to transmit all kinds of overhead bytes. The actual transmission service is only 7.2 Gb/s, and the transmission efficiency is only 75%, which means the efficiency is low. Third, the layer-4 architecture has excessive bandwidth granularity and overlapping functions. Layer-4 bandwidth allocation uses four completely different methods, namely, IP packets, ATM cells, SDH frames, and WDM wavelengths. In actual use, there is no need for so many bandwidth particles. In terms of functions, each layer has functions of adjacent layers, especially the protection and recovery functions. Each layer has functions that make it very complex and even conflict with each other. In short, the existing layer-4 network structure cannot meet the needs of data business development, and new technical means must be developed.

Figure 1 Development Trend of IP over WDM Networks

We know that MPLS Multi-Protocol Label Switching, which has developed rapidly in recent years, has been proved to be a technology that is very suitable for data transmission in electrical networks. MPLS uses the constraint-based routing technology to implement traffic engineering and fast rerunning, meeting service quality requirements. Therefore, MPLS-based routing technology can completely replace ATM in traffic engineering. Likewise, fast re-routing can replace SDH as a protection/recovery technique. It can be seen that using the Traffic Engineering provided by the IP/MPLS Control Platform and fast re-routing will allow the future transmission network to completely cross the ATM and SDH layers, as shown in Figure 1b )), directly implement IP over WDM. Undoubtedly, such an IP via MPLS over WDM network will be a network with simpler operations, the lowest cost, and the most suitable for data service transmission.
However, MPLS, after all, is a layer 2.5 Technology located between the Layer 3 network layer and Layer 2 in the OSI Layer 7 model, while WDM belongs to the optical layer and is the technology of the first physical layer. Therefore, to allow MPLS to directly act on the physical layer over the data link layer, it must be modified and expanded. Under such circumstances, the International Organization for Standardization (IETF) has promptly launched the common Multi-Protocol Label exchange technology (GMPLS) available for the optical layer.
To achieve seamless integration of IP and WDM, GMPLS expands MPLS labels so that labels can be used not only to mark traditional data packets, it can also mark TDM time slots, optical wavelengths, optical wavelength groups, optical fibers, etc. In order to make full use of the resources of WDM Optical Networks and meet the needs of future new businesses such as VPN and optical wavelength leasing ), to achieve intelligent optical network, GMPLS also modifies and supplements signaling and routing protocols. To solve various link management problems in optical networks, GMPLS designs a new link Management Protocol (LMPLink Management Protocol). To ensure the reliability of optical network operations, GMPLS also improves the protection and recovery mechanisms of optical networks. The characteristics of GMPLS are described below.
I. general multi-protocol labels
1.1 GMPLS Interface
We know that by adding a 32bit "shim" label to the IP Address Header, MPLS enables connection-oriented IP address transmission, which can greatly speed up IP packet forwarding. GMPLS expands the labels by uniformly marking TDM time slots, optical wavelengths, and optical fibers, so that GMPLS not only supports IP packets and ATM cells, moreover, it can support voice-oriented TDM networks and WDM optical networks that provide large-capacity transmission bandwidth, thus realizing the normalization mark of IP Address data exchange, TDM Circuit Switching mainly SDH) and WDM optical switching.
GMPLS defines five interface types to implement the above normalization labels:
(1) group switching interface PSCPacket Switch Capable): Performs group switching. The Group boundary is identified and forwarded to the group based on the information in the group header. For example, the MPLS label exchange router LSR forwards data based on the "shim" label;
(2) L2 switching interface L2SCLayer2 Switch Capable): Performs cell switching. Identify the boundaries of cells and forward the cells according to the information in the cell header. For example, an atm lsr is based on an atm vpi/VCI forwarding cell;
(3) Time Slot Switching interface TDMCTime Division Multiplexing Capable): Forward Services Based on TDM time slots. For example, electrical interfaces of sdh dxc devices can exchange SDH Frames Based on Time slots;
(4) wavelength switching interface LSCLambda Switch Capable): based on the optical wavelength or optical band forwarding service that carries the business. For example, OXC is a type of optical wavelength-level device that can be used to make forwarding decisions based on optical wavelength. Further, you can make forwarding decisions based on the optical band. Optical band switching is a further extension of optical wavelength switching. It regards a series of continuous optical wavelengths as a switching unit. The use of optical band switching can effectively reduce waveform distortion caused by single wavelength switching, reduce the number of optical switches, and reduce the interval between optical wavelengths .;
(5) optical fiber exchange interface FSCFiber Switch Capable): Based on the service optical fiber, it is forwarded to the actual location in the physical space. For example, an OXC device can connect one or more optical fibers;
The relationship between the above five GMPLS interface types is shown in 2.

Figure 2 GMPLS five Interface Types