Optimizing Optical Network with IP

Yan Xiaoying of Transmission Design Institute of Zhejiang Post and Telecommunications planning and Design Institute

There is no doubt that the revolution brought about by IP in the data communication world is decisive. It is conceivable that the first real handshake between the ITU-T and the IETF is tantamount to announcing the birth of a new generation of technology.

Based on the characteristics of metropolitan area Network data Services, there are several examples of optimizing optical networks with IP technology: RPR, Ion and IP over DWDM.

Compared with SDH, RPR uses a new reverse dual-loop mechanism for 50ms protection similar to the APS protocol, while the unique SRP technology saves half the bandwidth of SDH. As the current IEEE802.17 Working Group has not yet launched the final standard RPR specification, more manufacturers are unique to their own technology, to a large extent does not have the versatility of technology. Of course, the advantages of RPR is also obvious, such as the two-storey topology learning ability, COS byte mapping priority determination and routing convergence time reduction, the key is to solve the past GE can not be a ring problem. RPR uses the vanguard when counting luminous and Cisco. Luminous in January 2002 to assist Netcom in Shanghai, Dalian, Guangzhou and other nine cities in the construction of the city based on RPR, Cisco in Henan Telecom opened the RPR City area network, the domestic Huawei Company main push Ip-ring technology is a disguised RPR. In our view, however, the application of RPR in the short term is not promising. Operators, considering the existing SDH network, are more willing to use MSTP to extend their SDH. Even in a city where data traffic is very large, RPR is built on the edge of the network GSR, it is difficult to expect it to become the city's dominant technology.

With the proposed concept of ITU-T ASON, g.8080 finally succeeded in bringing everyone's eyes together. Ason defines three levels for the network: Control plane, management plane and transfer plane.

The control plane consists of a multitude of OCC, each of which corresponds to the actual device of a transport plane, provides call signaling control and routing connection selection for each optical node in the transport plane, and attempts to recover the connection by constantly monitoring and interacting with the LSA in an attempt to use protection switching in case of a network failure. In a sense, OCC completes some of the functions of the router, can continuously update the routing table and network topology structure. Now most of the OCC still need separate external equipment, but some manufacturers consider it as an interface card, directly plugged into the device's business slot.

Repeat construction is a relatively sensitive issue, operators of the network often appear such a problem: for a user to open a 2M line, after a long period of operation and the network adjustment of the multi-phase project, the protection of the 2M circuit has been increasing multiplicity. When users want to undo this 2M application, they will find that the protection circuit is already missing in the network management. This is tantamount to wasting a 2M bandwidth in disguise. It is easy to find and solve this problem by using the control plane of the GMPLS protocol.

The management plane has the entire network monitoring, the maintenance and the coordination function, the function is similar to the centralized network management of SDH, but the function is more powerful: including the best path calculated by the OCC to modify, the average distribution of traffic on the whole network, the automatic adjustment of end-to-end connection and different attributes of the business choice of different configurations and so on.

The nodes in the transfer plane are basically using DXC equipment because they want to realize a large number of services in Ason, and attach a fairly strong signal monitoring capability, and provide end-to-end physical layer connection directly to the user. For a small amount of management and control of the transmission of information, you can choose both in-band and Out-of-band two different ways.

Domestic manufacturers generally rarely produce pure DXC equipment, so the current Ason's main suppliers are limited to foreign countries such as Nortel, Sycamore and Ciena and other manufacturers. The construction process of Ason will determine the development of Oan, which is expected to be the best development opportunity of Ason in 2005 when the whole net adjusts period.

IP over DWDM is actually currently only as a development concept to promote, there is no practical applications and products, not even a mature technology. In fact, IP over DWDM is the most basic error is: IP is three layers, DWDM is a layer, it is impossible to throw open two layers to talk about three layers. Second, because the vast majority of the network is based on SDH, it is absolutely impossible to make a digital packet to do the frame of the network.

The ideal situation for IP over DWDM is of course oan. Because of the transparency of the business, we can not care about the second floor, but can never give up the second floor. So ultimately, IP over DWDM will still be limited to the two-tier frame structure of the business itself. In addition, the upper and lower wavelength business in the near future is not necessary, because the local business in the short term will still be low speed SDH signal, still need to undergo photoelectric conversion. Also, pure optical exchange of OXC and Oadm has not yet been able to scale commercial, destined to oan the arrival of a gradual process.

Another point to be mentioned is how to design the topology of the network. Data technology is very particular about the use of protocol for the whole network path calculation, it is recommended that the network topology using the FULL-MESH structure of the mesh network, to ensure that the multiple routes of the optional and node direct. However, the mesh network is difficult to be compatible with some of the original ring SDH network, coupled with the limitations of QoS, the pursuit of IP addressing may also embark on the old path of wasted bandwidth. Therefore, we recommend that the ideal metropolitan area network should be as far as possible to adopt a ring mesh structure. In this way, we not only retain the essence of the routing protocol, but also different types of protection for different COS requirements, and make full use of network resources.

Although the 50% bandwidth reserved by SDH Ring Network will cause some waste of resources, it can guarantee the QoS of the network well. In the same case, mesh to achieve this protection effect, the sacrifice of bandwidth is far more than 50%.

In a mesh-like network structure, perhaps we need only between A-b, C-d, e-f between the business, but in order to be able to guarantee the network COS in the higher SLA, we have to B-c, D-g, g-f and e-a bandwidth all occupied, in order to achieve a logical ring.

Of course, the use of pure Full-mesh structure can also be protected. But it is in the network failure of the time to protect the routing calculation, it is bound to increase a lot of delay time. And like SDH through reserved resources, the prior backup of the routing and bandwidth will greatly reduce network failure in the short term caused by the damage.