HK1062235A - Maintaining quality of packet traffic in optical network when a failure of an optical link occurs - Google Patents

Description

Maintaining quality of packet traffic within an optical network in the event of failure of an optical link

Technical Field

The present invention relates generally to supporting packet traffic within an optical network, and more particularly to protecting internet traffic in the event of a failure of an optical network link.

Background

Fig. 1 shows the OSI and TCP/IP communication models. The seven-layer OSI model was derived from the work of the standards committee, while four layers of TCP/IP built on top of the hardware layer resulted from practical studies. The session layer and presentation layer functions defined by the OSI model are omitted in the TCP/IP model, which functions are done by different TCP/IP protocols when needed.

In the TCP/IP model, users interact with network applications at the application layer. Data is received as instructions from the user or data from a network application connected to the other end. The TCP/IP application communicates within the client/service pair. The transport layer uses TCP (transmission control protocol) to manage the data flow between two internetwork hosts. At the network layer, data moves around the internet. The Internet Protocol (IP) operates at the network layer to route packets through the network independent of the network medium. The data link layer, also known as the network interface layer, is used to transmit data across a single network. The physical network includes several physical mediums: copper wire, optical fiber, radio channel.

The application layer and the transport layer function as an end-to-end protocol, and the protocol is related to communication between end systems. Instead, at the data link layer and the network layer, the protocol is related to the actual transmission path taken by the traffic. At the network layer, datagrams are addressed to the final source host, but intermediate routers will examine the destination address and route the traffic locally in any desired manner.

Local network addressing is important at the data link layer because it only knows the hardware address of the host over the same physical wires. Thus, the data link layer displays the source and destination addresses of one or more routers.

By assigning different functions to different network layers, traffic can be routed through networks (the internet) that span the globe. Only the intermediate router needs any significant amount of information about the in-network structure and the host only needs to know which services are local and which are not.

The reliability of TCP transmission is based on the use of reception acknowledgements, retransmission requests and the use of timeouts. IP transport does not provide any guarantees on transmission rate, bandwidth, delay, and throughput. In other words, the IP protocol is different from another widely used protocol, Asynchronous Transfer Mode (ATM), which does not provide any quality of service guarantees.

As previously mentioned, the Internet Protocol (IP) operates at the network layer to route packets through the network independently of the network medium, while the data link layer is used to transmit data across a single network, which may include several physical mediums.

High speed networks such as SONET (synchronous optical network) and SDH (synchronous digital hierarchy) use optical fiber as the physical transport medium. Today optical networks carry a large portion of internet traffic as part of the public telecommunication infrastructure.

The protection of the optical network against fiber cut-off will be explained briefly below. The basic principle of optical protection is to set up a backup path for traffic. The backup path implies that another fiber is routed to another fiber. Two basic protection principles are used for simple point-to-point links: 1+1 protection and 1: 1 protection.

Fig. 2A depicts 1+1 protection, where traffic is transmitted from a source to a destination simultaneously on two separate fibers. One fiber is the working fiber and the other fiber is the protection fiber, wherein the splitter transmits the same data to both fibers. Thus, in 1+1 protection, there are two fibers from the source to the destination, and traffic is transmitted simultaneously on two separate fibers. The switch selects one of the two optical fibers for reception. If the working fiber is severed, the destination switches to the protection fiber and continues to receive data. The switching time is rather short, about 2 ms.

FIG. 2B depicts 1: 1 protection. Traffic from the source is transmitted via only one fiber at a time, i.e. only via the working fiber. In normal operation, the other fiber, the protection fiber, is "cold"; no data is transmitted. In a unidirectional communication system, fiber cuts are detected by the destination rather than the source. Thus, if the working fiber fails, the destination will detect the failure and the optical switch switches to the protection fiber. The destination must then use a signaling protocol to switch to the protection fiber notification source. In two-way communications, a fiber cut will be detected by both the source and destination. In 1: 1 protection, optical switches at both ends of the link are required. The switching time is clearly greater than 1+1 protection.

IP routers are responsible for routing IP packets within the internet. Routers forward network traffic from one connecting network to another. Furthermore, the network may be an optical network, and there may also be several intermittent optical networks in between. Complicating the problem in routing IP packets through an optical network at an IP router is that IP networks and optical networks include many layers. Each layer within the two networks has its own protection. Furthermore, there is no interworking between the protection mechanisms of the optical network and the network layer of the internetwork. Thus, the network layer on which the Internet Protocol (IP) operates is completely independent of the optical layer of the optical network and, accordingly, independent of the protection of the optical fiber.

A disadvantage of the above features is that the IP router has no knowledge of how to establish the optical transport layer, i.e. the IP layer does not have full knowledge of the optical routes between the nodes. Therefore, the nature of the traffic transmitted via the optical network is not noticed when optical protection is set against fiber cuts. The disadvantages are more pronounced in terms of the quality of service (QoS) specified for internet transmissions. In the prior art, Internet Protocol (IP) treats the optical layer as a simple point-to-point connection without optimizing capacity usage to match the QoS level of the Internet Protocol (IP). On the other hand, the optical layer does not support QoS for internet protocols.

Disclosure of Invention

The invention aims to design a method capable of realizing the interworking of optical protection and IP layer protection so as to support QoS routing and IP packet forwarding.

The object is achieved by configuring the optical point-to-point links to form part of a ring network and setting different protection types for different links. By using 1+1 protection or 1: 1 protection for each optical link within the ring network, each of the optical links may have an appropriate level of protection corresponding to the nature of the internet traffic transmitted over the link.

Although a fiber cut may cause a link failure, the high protection level of the link almost guarantees uninterrupted transmission of internet traffic at the same bit rate as before the failure. The protection stage is implemented by means of 1+1 protection. The optical layer may provide such protection for high priority internet traffic that cannot tolerate delay.

Although a fiber cut may cause a link failure, the protective intermediate stages of the link ensure that internet traffic is transmitted after a short outage at the same bit rate as before the failure. The optical layer may provide such protection for high priority internet traffic that tolerates a few delays. The protection stage is implemented by means of 1: 1 protection.

The main difference between the high protection stage and the medium protection stage is its response time to a fault.

In the event of a link failure caused by a fiber cut, the low level of protection of the link does not guarantee continuous transmission of internet traffic. Therefore, no protection is provided within the optical layer. Upon loss of internet traffic, the IP layer will immediately detect the lost link, after which the routing table will be changed to accommodate the new situation. If the rest of the network is not congested, internet traffic using the lost link will be recovered.

In the medium protection level, optical signaling within the optical layer is responsible for protection, where the IP layer does not know when to perform protection actions.

The protection levels within the optical layer may have corresponding priorities within the IP layer.

Drawings

The invention will be described in more detail below with reference to the accompanying drawings, in which

FIG. 1 depicts a layer model of the TCP/IP protocol;

FIG. 2A illustrates 1+1 protection of an optical link;

FIG. 2B illustrates 1: 1 protection of an optical link;

FIG. 3 illustrates an optical ring network;

FIG. 4 depicts an arrangement from the perspective of the IP layer, an

FIG. 5 illustrates 1: N protection of an optical link.

Detailed Description

Fig. 3 shows a ring network comprising optical fibers as physical medium. The network includes two optical rings and three routers connected to the optical rings via Optical Interfaces (OIFs). In the event of a failure, the loop provides a high degree of availability while being topologically simple. Although a link may fail due to a fiber cut and a node may fail due to a power outage or equipment failure, the ring network is resilient to failure because it provides at least two independent paths between any pair of nodes. The path does not have any nodes or links in common, except for the source node and the destination node.

Referring to fig. 3, traffic between routers 1 and 2 is protected by 1+ 1: the optical fiber 21 is a working optical fiber and the optical fiber 23 is a protection optical fiber. Parallel fibers with traffic in the same direction may be replaced by a single fiber capable of carrying multiple optical channels. For example, Wavelength Division Multiplexing (WDM) techniques can be used for this purpose, wherein wavelengths can be added to and removed from the fiber by using WDM multiplexers and demultiplexers, respectively. Traffic between routers 1 and 2 and between routers 1 and 3 is bidirectional. For simplicity, only the traffic direction from router 1 to routers 2 and 3 is considered below.

Router 1 has an interface OIF 11 for transmitting data to router 2 and, correspondingly, router 2 has an interface OIF 21 for receiving data from router 1. The switch 220 monitors the optical power from the fiber 21 and if the optical power disappears due to fiber cut, the optical switch 220 simply switches to the fiber 23 and continues to receive data. The switching time is very short; approximately 2 ms. Only one optical interface is required at both ends. For example, if the total capacity of the optical fiber between router 1 and router 2 is 2.5Gbit/s, then only 2.5Gbit/s of capacity is provided for traffic between routers 1 and 2, not a maximum capacity of 5 Mbit/s. On the other hand, due to the 1+1 protection, the capacity is available not only under normal operating conditions, but also when effective protection occurs during fiber failure.

The protection provides a high level of protection for internet traffic. In most cases, switching to the spare fiber is so rapid that the IP layer is not aware at all that a failure has occurred in the optical layer. Thus, a point-to-point connection protected by 1+1 can be provided to users whose internet traffic requires a very reliable connection.

Intermediate and low protection levels may be provided for traffic between routers 1 and 3. There are two optical interfaces within routers 1 and 3: router 1 has interfaces OIF 12 and OIF 13 for transmitting data, and router 3 has interfaces OIF 31 and OIF 32 for receiving data from router 1, respectively. As in the previous example, there is also traffic from router 3 to router 1, but this is not shown in fig. 3. Thus, there are two optical links between routers 1 and 3. If the capacity of each link between router 1 and router 3 is 2.5Gbit/s, the maximum available capacity between the routers is 5Gbit/s, where both fibres are used for the traffic. In this case the load distribution principle is used for the transmitting router to distribute traffic between the optical interfaces OIF 12 and OIF 13.

According to the present invention, traffic between router 1 and router 3 is 1: 1 protected. According to the 1: 1 protection scheme, the optical fiber 24 of the link is selected as the "working fiber" and the optical fiber 22 is the "protection fiber". It should be noted that each fiber may be selected as the working fiber. For example, traffic transmitted by router 1 to fiber 24 via optical interface 12 and optical switch 210 is protected, and thus router 3 is always able to receive the traffic from fiber 24 via optical switch 230 and optical interface 32, or from fiber 22 via optical switch 230 and optical interface 31.

However, in contrast to the 1+1 protection philosophy in which the protection fiber is "cold", traffic is also transported via the protection fiber 22 in normal operation. Thus, the IP router 1 sends the packet to the optical fiber 24 through the optical interface OIF 13 and the optical switch 210. Packets with low priority are routed to the optical fiber 22 through the optical interface OIF 12 and the optical switch 210. The same bit rate will be provided for all packets transmitted between router 1 and router 3 regardless of their priority levels.

If fiber cleaving occurs within the optical fiber 24, then protection operations according to a 1: 1 scheme will be performed. Optical switch 230 detects that no packets have arrived from fiber 24, i.e., no light has arrived from fiber 24, then switch 230 switches to route packets from fiber 22 to optical interface 32 and from fiber 24 to optical interface 31. At the same time, switch 230 notifies switch 210 of the switch change using a signaling protocol, whereupon switch 210 switches to directing packets from optical interface 13 to optical fiber 22 and from optical interface 12 to optical fiber 24.

The result is that after a short break, medium priority packets are still transmitted from router 1 to router 3 via another fiber as before the failure. Packets with low priority are directed to the damaged fiber 24 and will be lost.

In the protection scheme described above, low priority packets are transmitted at the same bit rate as medium priority packets, but in the event of a failure, medium priority traffic survives while low priority traffic is interrupted. Thus, low priority traffic always experiences a risk of being dropped.

The optical switches 210 and 230 do not change their positions if fiber cleaving occurs within the optical fiber 22. As a result, medium priority traffic via fiber 24 survives, while low priority traffic via fiber 22 is interrupted.

In summary, medium priority packets can always be transmitted between routers 1 and 2 regardless of which link the fiber cut occurs in.

The router determines which packet will be routed to which optical interface. The determination is done without knowledge of the underlying optical network. In any case, the operator of the optical network presets the protection of the optical network and the optical fibers and configures the routers in a suitable way so that the routers route a certain traffic to a certain optical fiber that provides a certain priority in view of the traffic needs.

The classification of traffic as priority may be performed, for example, by the destination and/or source of the IP packet. After classification is complete, the type of protection between the appropriate nodes will be selected and the link will be configured accordingly. The router then directs the packet to the appropriate optical interface and further to the appropriate optical fiber. However, the configuration of the optical links within the ring network is substantially static and the set configuration changes little. In any case, the router determines how to direct traffic to the fiber.

It should be noted that the present invention combines protection within the IP layer with protection within the optical layer, although the layers are completely independent of each other. The quality of internet traffic between nodes is maintained despite the absence of any control signal flow between the optical layer and the IP layer. The existing internet protocol supports a 1+1 and 1: 1 protection scheme.

The optical protection switching according to the invention is particularly suitable for the internet with emerging quality of service (QoS) routes being developed. In QoS routing, the links between routers are associated with QoS parameters. Routing tables are generated for different transmission classes, respectively.

This will be described in more detail below with reference to fig. 4 and 3.

Fig. 4 shows routers 1 and 3 of fig. 3 and the links between them. In addition, fig. 4 also shows the arrangement from the perspective of the IP layer. Link 1 corresponds to fiber 24 and link 2 corresponds to fiber 22. Consider traffic from router 1 to router 3. The router 1 checks the QoS parameters of the incoming IP packet. If the parameters indicate that the packet requires high reliability and low delay, the packet is routed through link 1. This route is depicted as a solid arrow in fig. 3. Other packets, i.e. packets whose QoS parameters indicate that they tolerate longer delays and have low reliability requirements, are routed through link 2. This route is depicted as a dashed arrow in fig. 3. Packets traversing link 1 may be charged at a higher price than packets traversing link 2 in view of the higher QoS of link 1.

If a light cut occurs in link 2, the link is deleted during the repair period. If the IP network is not overloaded, protection at the IP layer will occur in a few seconds and the IP connection is restored. Link 1 also experiences only a very brief interruption, if any, and the failure does not trigger any protection at the IP layer.

Three different approaches may be used at the IP layer; a load distribution scheme, an improved QoS packet forwarding scheme, and a QoS routing scheme.

The load distribution scheme is used for current routers, but the scheme does not take advantage of the knowledge that link 1 survives the ray cut and link 2 does not survive.

Within the improved QoS packet forwarding scheme, if there is a possibility of link 1 blocking, low priority packets transmitted via link 1 are directed to link 2 rather than being dropped.

In a QoS routing scheme, the links have different routing parameters as described above.

Figure 5 depicts an optical network that allows five priorities. Fig. 5 differs from fig. 3 in that there are four optical fibers in the loop between router 1 and router 3. Of course, the number of additional links between routers is not limited to five, but any number N of fibers may be used. Thus, the protection scheme is referred to as 1: N. In the 1: N protection scheme, N working fibers share a single protection fiber, where protection can handle faults within any single working fiber. Thus, each of fibers 51, 52 and 54 may transmit high priority traffic between routers 1 and 3, while fiber 55 carries low priority traffic. If a fiber cut occurs within fibers 51-54, its traffic is routed to fiber 55 and the low priority traffic of fiber 55 is dropped.

Thus, each fiber 51-54 may be assigned a different priority. If a fibre cut occurs in the fibre 52 carrying the highest level of traffic, the traffic will be routed to the fibre 55 whose traffic will be dropped. If a failure occurs within fiber 51 thereafter, its traffic will be routed to fiber 53 having a lower priority than fiber 55 because fiber 55 is carrying higher priority traffic than fiber 51. Further, the total capacity of the router output may be divided between each optical interface OIF 12.. OIF N and OIF 32.. OIF N. The typical capacity of current optical interfaces is 2.5 Gbit/s. Traffic with a rate of 10Gbit/s may then be distributed between the 5 links 51-55.

If a 1: N protection scheme is used, the same number of priorities may be required within the global IP.

The invention is applicable to ring networks, and in particular to Metropolitan Area Networks (MANs) and SONET/SDH networks.

The proposed method is suitable for billing clients, for example. Thus, charging may be based on the QoS required by the client, rather than on the traffic as in the prior art.

Claims (10)

1. A method of protecting packet traffic from failure in an optical network comprising routers, optical fibers, and optical switches interconnecting the routers and the optical fibers, the method comprising the steps of:

setting 1+1 protection including two optical links in an independent optical fiber for high-priority packet service between a transmitting router and a corresponding receiving router;

routing said high priority packets to said two optical links at said transmitting router, wherein after a fiber cut occurs in one of said optical links, said corresponding receiving router continues to receive said packets from the remaining optical links without significant delay; and

setting up 1: 1 protection of at least a first optical link and a second optical link in an independent optical fiber for medium-priority and low-priority packet services between a transmitting router and a corresponding receiving router;

routing the medium priority packets at the transmitting router to the first optical link,

routing low priority packets at the transmitting router to the second optical link, an

Rerouting said medium priority packets at said transmitting router to said second optical link and said low priority packets to said first optical link in response to a fiber cut in said first optical link, whereby said medium priority packets continue to be received at said corresponding receiving router after a short handoff delay but said low priority packets are lost;

in response to a fiber cut within the second optical link, routing is maintained at the transmitting router so that the medium priority packets continue to be received at the corresponding receiving router without delay, but the low priority packets are lost.

2. The method of claim 1, wherein the optical fiber and the router are coupled as a bidirectional loop having at least two optical links between the transmitting router and a corresponding receiving router.

3. The method of claim 1 or 2, further comprising the step of:

assigning different quality of service parameters to high, medium and low priority optical links at the IP layer of the Internet protocol, and

enabling the router to generate different routing tables for different quality of service classes.

4. The method of claim 1, further comprising the steps of:

performing rerouting by changing the state of optical switches at the transmitting router and the corresponding receiving router, wherein protection at the optical layer is completely independent of protection at the IP layer.

5. The method of claim 4, wherein the transmitting router routes packets to the appropriate optical interfaces according to their priorities and continues the same route during a failure in any of the optical links.

6. A method according to claim 1 or 4, wherein a parameter defining the quality of service is appended to each of said packets, and said transmitting router routes said packets to the appropriate optical interface in dependence on said parameter.

7. The method of claim 1, further comprising the steps of:

further routing said low priority packets to said first optical link, wherein said low priority packets are transmitted with medium priority packets;

directing said low priority packets to said second optical link in response to packet congestion within said first optical link.

8. The method of claim 1, further comprising the steps of:

setting 1: N protection of 1+ N links including optical fibers for N +1 priority traffic between a transmitting router and a corresponding receiving router, wherein each optical link transmits packets of different priorities;

in response to a fiber cut within any one of the fiber links, rerouting packets for that fiber link to the link carrying the lowest priority packet, an

Discarding the lowest priority packet.

9. The method of claim 1, further comprising the steps of:

only one optical link is provided for a single optical fiber, wherein the number of links is equal to the number of optical fibers.

10. The method of claim 1, further comprising the steps of:

a plurality of optical links is provided for a single optical fibre by using wavelength division multiplexing techniques, the number of said links being greater than the number of said optical fibres.