WO2013010593A1

WO2013010593A1 - Data traffic distribution in an ethernet node

Info

Publication number: WO2013010593A1
Application number: PCT/EP2011/062556
Authority: WO
Inventors: Juergen Bos; Adrian Michel
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2011-07-21
Filing date: 2011-07-21
Publication date: 2013-01-24

Abstract

The invention relates to techniques for controlling data traffic distribution in an Ethernet ring node being part of an Ethernet ring. It is proposed to administrate assignment entries with an assignment between a ring port and a host in the Ethernet ring being reachable over said ring port. Further the assignment entry comprises a distance information providing information on a distance between the ring port and the host. In case of a topology change it is proposed to determine any change in a reachability of the host over the ring port after the topology change. The determination is based on the distance information and a location of the change point, wherein the location of the change point is provided after detecting a topology change. Based on the determination result, an updating procedure on the assignment entry is performed considering the location of the change point if the reachability of the host has changed

Description

Data traffic distribution in an Ethernet node

Technical Field

The present invention relates to an Ethernet ring node distributing data traffic in a Ethernet ring and it relates to a method for controlling the data traffic distribution in the Ethernet ring node.

Background

Carrier Ethernet Networks have - amongst other requirements - the requirement to provide high availability. This is reached by avoiding outage times of the Carrier Ethernet Network Elements and Components as well as by network protection mechanisms against link failures and node failures. In the latter case, fast restoration times are necessary to be able to run Ethernet as Carrier Ethernet. The ΓΓΙΙ-Τ G.8032 standard for fast Ethernet Ring protection promises < 50 msec failover times on link level in certain situations.

A G.8032 Ethernet ring based on the ITU-T G.8032 standard has one ring protection link (RPL), which is blocked in normal operation, thus avoiding loops. In normal operation, the basic idea is to use the Ethernet Ring Protection (ERP) connecting two

Ethernet nodes to block the data traffic on this link in order to create a loop free structure. In case of a failure on a link or port, signal failure (SF) messages are flooded to inform other ring nodes of the failure condition. On protection switching, the RPL is unblocked and a new RPL connected to the failure is created forming a new topology and thus a new traffic pattern on the ring. This change on the ring is called topology change. An Ethernet ring node is adapted to connect the adjacent Ethernet nodes over corresponding ports assigned therefore. The Ethernet node is further adapted to connect hosts or clients over corresponding ports. The hosts are reachable via address information, e.g. the MAC address. A Filtering Data Base FDB - also called Forwarding Data Base - of an Ethernet node keeps track of which MAC addresses are learned on which interface ports. In other words during the learning phase the FDB gets information which host can be reached over which port, wherein the reachability of the hosts on the Ethernet ring considers the location of the RPL. Thus, from a particular port hosts being located before the RPL can be reached. The hosts may be attached directly to the Ethernet node or they may be reached via the other Ethernet node being provided in the Ethernet ring.

The FDB is dynamic in a sense that if new MAC addresses are learned on the ring ports the FDB is updated correspondingly. This MAC learning process resulting in the assignment entries between the ports and the reachable hosts over said ports provided in the FDB. The provision of the assignment entries allows an Ethernet node to not just flood every incoming packet on all outgoing ports but selectively send packets to a sub-set of interfaces.

Fig. 1 shows an example of an Ethernet ring, 10. The Ethernet nodes A to G are connected via the Ethernet ring, 10. Each Ethernet node has two ports rpl and rpO. To some Ethernet nodes, hosts are connected; hosts HA1 and HA2 are connected to the Ethernet node A; hosts HB1 and HB2 to node B; HD to C; HD to D and HD to E. Further the RPL is depicted between ring node A and ring node G.

According to the G.8032 standard each of the ring nodes administrates a FDB. This administration is presented in respect to Fig.2 showing a FDB of the ring node B.

On the left side of the Fig.2 the ports are listed and on the right side the address information of the hosts, MAC address. Over port rpl the hosts HC, HD and HE and over port rpO the hosts HA1 and HA2 can be reached. Thus the Ethernet ring node B with the aid of the FDB sends the received data traffic towards the host to which the data is foreseen via the corresponding port. The knowledge of the assignment between the ports and the nodes is achieved during the learning process of the MAC addresses.

When a topology change is detected, all Ethernet ring nodes are supposed to remove their learned MAC addresses from their FDBs. This procedure is called FDB flush. From this point on, a procedure for learning the MAC addresses of the attached hosts and clients is started in all ring nodes by flooding data frames to the Ethernet ring until MAC address learning process is finished. This however prolongs the restoration time and leads to additional traffic which has a negative impact on frame losses and delay. The FDB flush method as described in ITU-T G.8032 standard has namely the disadvantage of a large amount of transient traffic overshoot caused by flooded Ethernet frames right after protection switching has taken place. This traffic overshooting even becomes more critical when a ring provides services to a large number of clients, then this implies a large number of MAC addresses to be learnt.

Taking the Fig.l and assuming that a failure has occurred on the link between the ring nodes C and D, shown as an arrow, as a consequence the RPL between the A and G ring node is to be unblocked changing the topology of the ring.

Returning to Fig.2, in case of the topology change, some entries are become invalid. For example it is not possible anymore to reach host HD and HE over port rpl of node B. Therefore in the existing solutions it is proposed to perform the FDB flush for all the MAC entries that are related to the ring ports and to start the MAC learning process from scratch.

This has however the effect that all traffic from/to the ring ports will be flooded over both rings ports, leading to the data traffic overshooting. Further the learning of the MAC addresses of the hosts being reachable in the Ethernet ring leads to a long recovery time before the Ethernet ring is working again. Summary

There is a demand for a technique for controlling the data traffic distribution in case of a topology change in an efficient way.

The invention is embodied in independent claims. Advantageous embodiments are described in the dependent claims.

The demand is satisfied with a method for controlling data traffic distribution in an Ethernet ring node being part of an Ethernet ring. The Ethernet ring node is configured to connect at least one host and comprising two ring ports, each ring port adapted to connect an adjacent Ethernet ring node. Further the Ethernet ring node comprises an assignment entry with an assignment between each ring port and a further host in the Ethernet ring being reachable over said ring port. The method comprises providing distance information including information on a distance between the ring port and the host. It is proposed that the distance information is included in the assignment entry. Further the method comprises detecting a topology change of the Ethernet ring by detecting a change point in the distribution of the data traffic in the Ethernet ring. It is proposed that the detection is based on the reception of information providing the location of the change point. In the next step any change in a reachability of the host over the ring port after the topology change is determined, wherein the determination is based on the distance information and the location of the change point. In case the determination result is that the reachability of the host has changed, it is proposed to perform an updating procedure on the assignment entry considering the location of the change point.

The abovementioned demand is also satisfied by an Ethernet ring node adapted to control data traffic distribution and being part of an Ethernet ring. The Ethernet ring node comprises two ring ports, each ring port adapted to connect an adjacent Ethernet ring node. The Ethernet ring node comprises also at least one further port being configured to connect at least one host. Further the Ethernet ring node comprises an as- signment entry being adapted to provide an assignment between each ring port and a further host in the Ethernet ring being reachable over said ring port. Further the assignment entry comprises a distance information providing information on a number of Ethernet nodes between the ring port and the host. Further the Ethernet ring node comprises a detector adapted to detect a topology change of the Ethernet ring by detecting a change point in the data traffic in the Ethernet ring, wherein the detection is based on the reception of information providing the location of the change point. A control unit in the Ethernet ring node is adapted to determine a reachability of the host over each ring port after the topology change wherein the examination is based on the distance information and the location of the change point. A processor being connected to the control unit and to the assignment entry and being adapted to perform based on the examination result an updating procedure on the assignment entry considering the location of the change point if the reachability of the host has changed.

The advantage of the solution is that no flush on the assignment entries is needed any longer and the ring protection works faster and without traffic overshooting.

Brief Description of the Drawings

In the following, the invention will further be described with reference to exemplary embodiments illustrated in the figures, in which:

Fig. 1 schematically illustrates an Ethernet ring according to state of the art;

Fig. 2 illustrates assignment entries in a database according to state of the art; Fig.3 is a flow diagram exemplarily illustrating an operation of the embodiment of the invention performed in the Ethernet ring node;

Fig.4 schematically illustrates components of the Ethernet ring node;

Fig.5 schematically illustrates provision of the assignment entries Ethernet ring node;

Fig.6 illustrates assignment entries in a database according to an embodiment of the present invention;

Fig.7 schematically illustrates an embodiment of ring size evaluation;

Fig.8 schematically illustrates an embodiment of change point detection;

Fig. 9 illustrates an update of the assignment entries in a database according to an embodiment of the present invention.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular network environments and communication standards etc., in order to provide a thorough understanding of the current invention. It will be apparent to one skilled in the art that the current invention may be practiced in other embodiments that depart from these specific details. For example, the skilled artisan will appreciate that the current invention may be practised with any network having a ring structure.

In the following the description is based on the G.8032 standard. However this should not be seen as a limitation to said standard, since the invention may be implemented in other networks having a ring topology.

The Ethernet ring node may be realized in any suitable and preferably way. In the following as an example the Ethernet ring node based on the G.8032 standard is used for presenting some features of the invention. An Ethernet ring node based on said standard is also called G.8032 node. An Ethernet ring node has got a plurality of different ports. There are at least two ring ports being configured to connect an adjacent Ethernet ring node. The Ethernet ring node may comprise further ports being adapted to connect a host or any type of clients to said Ethernet ring node. Moreover, a further network with attached hosts may be connected to a Ethernet ring node, like for example a LAN with a plurality of hosts. In the following description said ports will be called host ports and with the term "host" also the possibly attached networks or any kind of clients are meant.

The change point describes any change in the distribution of the data traffic. In particular the change point may be connected to any failure which may occur in the Ethernet ring. The failure may for example arise on a link between two adjacent ring nodes, so that no data may be sent over the failed link. In this context it is to be mentioned that a link failure may also occur on the RPL link. In a further example a failure may occur in a ring node, in this case the change point is the ring node. An Ethernet ring node may fail with the consequence that no data can be sent through said node. Moreover, the change point may be also a link or a node during the recovery phase, since it takes some time until the failure is fixed and the Ethernet ring works in a normal way. In this phase also no data distribution is possible. Further example may be a double failure, when for example a link and a node fail. Summarizing, the change point is connected to any failure which occurs in the Ethernet ring and which has an impact on the topology change of the Ethernet ring.

The host identifier describes an identifier for reaching a host in the Ethernet ring. Preferably the host identifier is an address being adapted to provide information for forwarding data frames to a host with said address. In one embodiment it is proposed to use the MAC address as a host identifier. The Media Access Control address (MAC address) is a unique identifier assigned to network interfaces for communications on the link layer in numerous network technologies including Ethernet. However there may be other possibilities of addressing. For example in case a further network, like a LAN with the attached hosts is connected to a Ethernet ring node, then an appropriate and suitable way of addressing is to be used to reach a host being attached to a LAN, for example by using additionally to the MAC address of the host also the address identifying the LAN.

According to an embodiment, the Ethernet node is configured to connect the hosts to the Ethernet node by means of a host port. However it is not mandatory that a ring node has a host being attached. It may be that the Ethernet ring node does not have any hosts being attached.

According to an embodiment, the assignment entry may be administrated in a database. Preferably, the assignment entries are administrated in a Filtering Database, also called Forwarding Database, FDB. The term assignment entry should not be understand as one entry, since it provides an assignment between each ring port and a further host in the Ethernet ring being reachable over said ring port. Thus there may be a number of entries for each of the ring ports and if there is a host being attached also for said host.

According to an embodiment, the distance information is determined from a host distance counter received on the ring port, wherein the distance counter provides a number of passed Ethernet ring nodes between the ring port and the host. Preferably the host distance counter is initiated by the Ethernet node to which said host is connected. Upon reception of the host distance counter, the Ethernet ring node stores the distance information in the assignment entry. The assignment entry comprises an assignment of the ring port and the host identifier together with the distance between said port and the host. In the next step, the host counter identifier is increased and forwarded to the next ring node. The updating and forwarding of the host distance counter is performed preferably during the phase when the Ethernet ring nodes learn about the hosts attached to the Ethernet ring. After the learning phase, the distance information between the ports and the host is provided in the assignment entries in the corresponding Ethernet ring nodes.

According to an embodiment, the location of the change point is determined from a change point distance counter received on the ring port wherein the distance counter provides a number of passed Ethernet ring nodes between the change point and the ring port. Preferably the change point distance counter is initiated by the Ethernet ring node being adjacent to the change point. In case a failure occurs on the link between two Ethernet ring nodes, preferably both nodes initiate and send the change point distance counter. On the way from the Ethernet ring node initiating the change point distance counter each Ethernet ring node which receives said counter increases the counter and forwards it further. At the end each of the passed ring nodes has the information about the location of the change point.

According to an embodiment, the determination step is based on the comparison of the change point distance counter and the host distance counter. In case when the change point distance counter is higher then a particular host distance counter it may be assumed that the host is further located than the change point, with the consequence that the host is not reachable anymore over the ring port assigned to said host.

According to one embodiment it is proposed that in the frame of the updating procedure a changing of the assignment between the ring port and the host is performed. Thus in case when the result of the examination procedure is that the host is not reachable from a particular ring port because the host is located behind the change point, then it is proposed to move the host to the other ring port.

Further it is proposed to provide a size of the Ethernet ring by using a ring size counter received on the ring port. The size of the Ethernet ring provides the number of

Ethernet ring nodes being connected in the Ethernet ring. The size of the Ethernet ring may be determined in any suitable and preferably way. For example it may be a static value being known to all the Ethernet ring nodes in the Ethernet ring. In other embodiments it is proposed to determine the ring size dynamically. In one example it is proposed that an Ethernet ring node being an owner of the ring protection link RPL initiates and sends on both ports an RPL distance counter. An Ethernet ring node when receiving said counter on both ports may determine the size by adding both values. In a further example, the ring size may be determined by sending a message with a size distance counter over the Ethernet ring. Each of the passed Ethernet ring nodes incre- merits the value of the counter and forwards it to the next Ethernet ring node. The Ethernet ring which initiated the message waits for the reception of said message with the determined size value. In the following step said Ethernet node propagates the information about the ring size to the other Ethernet ring nodes attached to the Ethernet ring. The advantage of the dynamical determination of the ring size is that when a new Ethernet ring node is added to the Ethernet ring a dynamic determination of the ring size is achieved.

Further it is proposed that the updating procedure comprises updating the distance information in the assignment entry by using the stored distance information and the ring size. Thus the new distance information may be calculated while taking the ring size minus the stored distance information. However there may be other ways of realising it, for example a message may be sent from the new port to the host for receiving the new distance.

Furthermore, the updating procedure may comprise deleting the assignment of the host to the ring port. In particular in case a failure has occurred in an Ethernet node, the hosts attached to said node can not be reached anymore. Thus, said nodes are to be deleted in the assignment entry.

It is proposed that the host distance counter HDC or the size ring counter are integrated in a data frame of the data traffic or in a signalling message. The distribution of the change point distance counter CP-DC is to be done by means of data frames. The corresponding counter may be integrated in a header of a data frame or in the pay load field. In case of the signalling solution, the signalling message used for transporting of the corresponding counter may be an existing message for example the R-APS message, wherein the counter may be integrated in any suitable and preferable field of said message. Further a new protocol message may be provided adapted to distribute the distance counter information.

With the proposed solution it is achieved that traffic flooding on all ports in case of failures is avoided. Further, when a topology change occurs, no extensive exchange of MAC address lists between the ring nodes is needed. Further due to the fact that the MAC addresses of the hosts on all ring ports are not to be learnt after a topology change, the service restoration time for the Ethernet ring after a failure is improved.

In the following an embodiment of the present invention is presented in respect to Fig. 3 showing a flow chart with steps according to the present invention.

While being in a working state, S31, the Ethernet ring node is adapted to provide information via which ring port a particular host is reachable. As a consequence, each Ethernet ring node has got an assignment entry with an assignment between the ring port and the host being reachable over the Ethernet ring and additionally with the distance information between said ring node and said host, S32. Herein the distance information means how many hops away from the ring port a host is located. Preferably the hosts are the Ethernet ring nodes which are to be passed to get from the ring port to the host.

While having the assignment entries, an Ethernet ring node is able to distribute data traffic according the host identifier in the assignment entries. In this state the Ethernet ring node may wait for detecting a topology change, S33. The topology change of the Ethernet ring is realised by receiving information providing the location of the change point, S34.

Upon detection of the topology change, the ring ports in the assignment entry are examined, S35, for determining any change in a reachability of the host over said ring ports, S36. The determination procedure is based on the distance information and the location of the change point. In one embodiment it is proposed to compare both values for each ring port and in case the change point distance counter is higher then the host distance counter it may be assumed that the host is located behind the change point with the consequence that said host is not reachable anymore over the ring port. Based on the determination step, a decision is taken whether an update on the assignment entry is needed. In case an update is necessary, the procedure goes to step S37 otherwise the entries are not updated and the procedure continue with the stored entries, S32.

In case an updating procedure on the assignment entry is needed, S36, a corresponding update is performed considering the location of the change point. The aim of the updating procedure is to move hosts to the second ring port if they are reachable now over the second ring port. Further the updating procedure comprises an update of the distance information for the hosts or/and a deleting of hosts if they are not reachable anymore.

In the following the Ethernet ring node adapted to perform the present invention is described with reference to Fig.4.

Fig.4 depicts an Ethernet ring node, 100, comprising a plurality of ports, 110, 120, 130. The ports are adapted to connect at least one host, 130. In this context, the ports are called host ports. Further there are two ring ports, 110, 120 being adapted to connect the adjacent Ethernet ring nodes.

Further, the Ethernet ring node comprises an assignment entry, 140. Said assignment entry provides an assignment between a ring port and a host being reachable over said ring port and distance information between said port and said host, wherein the distance information includes information about a number of Ethernet ring nodes, which are to be passed between the ring port and the host. In Fig.4 the assignment entry for one ring port is depicted. However the Ethernet ring node 100 may be adapted to administrate a number of assignment entry in accordance to the plurality of the ports

Further the Ethernet ring node 100 comprises a detector, 150, adapted to detect a topology change of the Ethernet ring by detecting a change point in the data traffic in the Ethernet ring. The detection is based on the reception of information providing the location of the change point, 180. The detector is connected to a control unit, 160. Said control unit has the task to examine the ring ports after the topology change in order to determine a change in the reachability of the hosts over the ring port. The determination of the change in the reachability of the hosts is based on the distance information and the location of the change point.

The result of the determination procedure is provided to the processor, 170. Said processor, 170 is adapted to perform an updating procedure on the assignment entry considering the location of the change point if the reachability of the host has changed. In order to perform the update, the processor is connected to the assignment entry.

In the following some embodiments of the present invention are presented.

In Fig.5, the embodiment for administrating the assignment entries is described; in particular the provision of the entries for the Ethernet ring node B is presented. Fig.5 shows an Ethernet ring, 10 as described in connection with Fig.l. The Ethernet nodes A to G are connected to the Ethernet ring, 10. Each Ethernet node has two ring ports rpl and rpO. Further there are hosts, HA1, HA2, HB1, HB2, HC, HD and HE connected to the corresponding Ethernet ring node via a corresponding host port, hpl or/and hp2. Preferably, the assignment entries are administrated in a Filtering Database, also called Forwarding Database, FDB.

In the following the provision of the assignment entries in an database, like the FDB, in the Ethernet ring node B is described in more details.

According to Fig.5 the host HD sends data to the host HB1. During the data traffic distribution, the Ethernet nodes B and C updates their assignment entries related to the host HD.

In step 51, the host HD sends a packet to node D. The packet may be either an Ethernet frame or it may be also realised by means of a special signalling packet. Upon reception of the packet, the Ethernet ring node D generates a host distance counter, HDC and adds said counter to the data packet when sending said packet out on the ring ports rpO, 53. Since the packet was received from a directly connected host, the node D sets the host distance counter, HDC to 1, 52. Further, the MAC address of the host HD is added to the data packet.

In the next step 53, the Ethernet node C receives the data packet and reads the MAC address with the corresponding host distance counter HDC out of the received packet. The node C updates its FDB entries by storing, 56, the MAC address of the HD host received on the port rpl together with the received distance information, namely 1. Before sending the Ethernet frame further on port rpO, 57 to the adjacent Ethernet node B, the node C increments the host distance counter HDC to the value 2, 55. Upon reception of the data packet on port rpl, 57, the Ethernet ring node B reads the entries in said packet and updates its FDB entries by storing the host identifier, MAC address, of the host HD and by setting the distance information to the host HD with 2, 58. Finally, the node B forwards the data packet to the host HBl after preferably deleting the HDC entry from the data packet, 59.

Fig.5 shows a learning process for the provision of the assignment entries corresponding to the host HD. This learning procedure is to be performed for both ports in each Ethernet ring node in the Ethernet ring, so that at the end each port in each node provides information in an assignment entry on the reachable hosts and the distance to said hosts.

Fig.6 shows an embodiment of the FDB entries for the node B. The FDB of node B in Fig.6 depicts the corresponding MAC addresses learnt on the corresponding ring port together with the corresponding distance information. On the left side of Fig.6 the ports, ring ports and host ports, are listed and on the right side the address information of the hosts, MAC address, together with the distance counter, HDC.

According to Fig.6, over the ring port rpl the hosts HC with 1 hop, HD with 2 hops and HE with 3 hops may be reached. Further the hosts HA1 and HA2 are reachable via one hop through the port rpO. The hosts HBl and HB2 do not have any entries; that means that they are reachable via 0 hops or in other words they are directly connected to the node B.

In one embodiment it is proposed to provide the ring size. Said size is used for calculating the distance information of the hosts which have been moved to another ring port in the Ethernet node after topology change.

In the following an embodiment for determining the ring size is presented according to Fig.7, which shows the Ethernet ring 10 as described in Fig.5.

According to the G.8032 standard, one of the Ethernet ring node adjacent to an RPL is assigned as the RPL owner. According to Fig.7, the node A fulfils the functionality of the RPL owner. Furthermore, the G.8032 standard provides a signalling message, R- APS. Said signalling message is used for propagating any information to the other nodes in the Ethernet ring. For example if a failure on a link or port occurs, the R-APS (SF) signal failure messages are flooded into the Ethernet ring to inform other ring nodes of the failure condition. The RPL owner propagates regularly the R-APS (NR, RB) message to inform the other ring nodes that the RPL is still alive. In a preferred embodiment it is proposed to use said signalling message to add the RPL distance counter, RPL-DC as an embodiment of the ring size counter. Thus, the format of the message may be, R-APS(NR, RB, <RPL-DC>).

According to Fig.7, the RPL owner, the node A initiates the RPL-DC with 1 meaning that the RPL is reachable with one hop, namely the host A. As next, the node A adds the ring size counter to the R-APS message, forming the message R-APS(NR, RB, 1). In the next step, the node A propagates on both ports, the R-APS(NR, RB, 1) messages to the adjacent Ethernet nodes, 701, 711. The messages pass the Ethernet nodes on the ring, 702-707 and 712-717, wherein each of the nodes increments the ring size counter, RPL-DC by 1 before forwarding it further. Upon reception of said message each node may determines the ring size. For example, taking the node B, it receives the RPL distance counter 1 on port rpO, 701 and 6 on port rpl, 716 and by adding of the two distance counters, the ring size, namely 7 is determined. It is proposed that each G.8032 node stores the ring size for calculating the new distance information for a host which is to be updated.

The determination of the ring size should be seen as an example. There may be another ways of providing the ring size. For example the RPL owner may send one R- APS(NR, RB, 1) message over one port, 701. Each of the passed nodes increments the RPL distance counter by 1. After receiving the response with the calculated RPL-DC on the other port, 707, the node A has got the ring size. In the next step, the node A may propagate the ring size result with a separate signalling message to the other Ethernet ring nodes.

In the following the behaviour of the nodes after detecting a topology change is described in respect to Fig.8, showing an Ethernet ring, 10 as described in connection with Fig.5.

Assuming that a link failure has occurred on the link between the ring nodes C and D, depicted in Fig.8 as an arrow, in this case the RPL between the nodes A and G is unblocked. As a consequence, the topology of the ring has changed and it is to be examined whether the assignment entries in the Ethernet ring nodes are to be updated.

When a link failure occurs, the nodes that experience the link failure send a R-APS(SF) message. In Fig.8, the node D sends on the port rpl, 811 and the node C on the port rpl, 801 the message R-APS(SF,1). Hereby, said nodes add to the generated R- APS(SF) message the change point distance counter, CP-DC, initiated with 1. Each ring node increases the value of the distance counter by 1. With the CP distance counter in the R-APS(SF) message, every ring node receives information on how many hops away the failed link is located. Taking the example of ring node B, this node receives distance counter 1 via ring port rpl, 801 and distance counter 5 via ring port rpO, 815. As a result of this embodiment, each Ethernet ring node has got the information about the failure in the ring, in this case about the link failure. Additionally the ring nodes are informed about the location of the failure in terms of the change point distance infor- mation providing the number of hops between a particular ring port in the node and the change point.

In the next step, the ring ports in the node B are examined for determining a reachability of the host over said port after the topology change, wherein the examination is based on the distance information and the location of the change point. Herein the stored and the newly received distance counters are compared, in particular the change point distance counter, CP-DC and the stored host distance counter, H-DC. Taking the example of Fig.8 with the link failure between ring node C and D and the entries according to Fig. 6, the examination on port rpl of node B comes to the result that the stored HDC for the host HD is higher than the received CP-distance counter, then 2 as a distance information assigned to the HD is higher then the received distance information to the link failure, 801, which is 1. That means for the ring node B that the link failure is closer than the host, or in other words the host HD can not be reached form the port rpl in the node B, since there is a link failure on the way thereto. The same result is achieved for the port rpl and the host HE, then the host distance information is 3 and the CP-distance counter is 1. Furthermore, the examination would come to the result that the host, HC on the port rpl is still reachable and also the hosts HA1 and HA2 on port rpO.

Based on the examination result, an updating procedure on the assignment entry is performed if the reachability of the host has changed. In the following the updating procedure is presented in respect to Fig.9 showing the FDB entries as described in connection with Fig.6.

Since in the examination step has recognised that the hosts HD and HE are not reachable via the ring port rpl, they are shifted to the other ring port, rpO. In Fig.9 this is shown by moving the entries from the port rpl to rpO and by removing the invalid entries under port rpl. However before removing the entries the distance information between the shifted node and the new assigned ring port is to be calculated. This may be performed in any suitable and preferably way. In one embodiment it is proposed to calculate the new distance information by using the ring size. Thus, the new distance information equals to ring size minus the stored host distance counter, HDC. Applying this rule to the example of Fig.9, this would mean that the distance from the port rpO to the shifted host HD would be 5, since the stored ring size is 7 and the stored HDC was 2. For the host HE, this means that the new calculated distance information is 4 as a result of the ring size 7 minus the stored HDC value, which is 3.

As a result of the update, the FDB has got new assignment of the entries. Thus, over port rp 1 only the host HC is reachable by one hop. The hosts HA1, HA2, HD and HE are reachable now via port rpO using the new calculated distance information.

In one embodiment it is proposed not to wait for the CP-distance counters to be received on both ring ports but to calculate the distance to the change point based on the ring size. Thus, if the first R-APS(SF,<CP-DC>) is received on any of the two ring ports, the distance counter from the other port can be assumed as ring size minus the already received distance counter. The advantage is that ring node may react quickly on the topology change. Preferably it is proposed to validate the assignment entry determined with only one counter when receiving the second counter to ensure that the updated assignments are correct. It may for example happen, that a double failure occurs in the network and that some nodes are not reachable via both ring ports. Further with both counters the determination can be made whether the failure has occurred in an Ethernet ring node or on the link.

As already mentioned, the change point may be a link failure, a node failure, a RPL failure or a link or node recovery. The aforementioned embodiments are based on the link failure. In the following the other embodiments for a change point are discussed.

In one embodiment the change point may be due to a failure in a node. Taking the embodiment according to Fig.8 and assuming that for example the node D of Fig.8 is affected by an failure. In this case the adjacent ring node to the failure are the ring nodes C and E. Said nodes generate the R-APS(SF,<CP-DC>) message which is then forwarded hop by hop through the Ethernet ring informing the ring nodes about the location of the failure. Upon reception of the CP-DC integrated in the message, for example the node B would detect that the host HD is not reachable anymore, so that the assignment entry leading to the hosts of the failed node is not valid. In the frame of the updating procedure, this entry is to be either deleted or set as invalid then there is no other way to reach said hosts, also not over the other ring port.

A further example of a change point may be a RPL failure. The RPL failure actually does not lead to a topology change since the RPL is blocked regardless of its state, errorless or failed. Upon reception of the R-APS(SF) message with the CP-DC, the nodes examine the ring ports and come to the result that all host are reachable with the already saved distance information, thus none of the assignment entries, (MAC address, Host Distance Counter) need to be changed.

In a further embodiment of a change point, the scenario of link or node restoration or recovery is considered. In this case each ring node is in pending state and receives R- APS(NR,RB, CP-DC) messages because of a recovery, wherein the R-APS(NR,RB, CP- DC) message is propagated from the nodes adjacent to the failed link or node. Preferably each time when this message is received the Ethernet ring nodes examine its entries and checks whether an update is necessary by comparing the CP distance counter with the stored distance information.

As already mentioned the propagation of the different counters, host distance counter HDC, the ring size counter, change point distance counter CP-DC may be realised in any suitable and preferably way.

In one embodiment it is proposed to include the information in the frames crossing the Ethernet ring during the transmission of data traffic. There are different possibilities for realisation of the integration of any of the distance counter in an Ethernet frame.

Typically the Ethernet ring is administrated by one authority and local changes do not affect the forwarding in the network. Depending on the implementation it is also possi- ble to transparently carry modified frames through nodes that are not distance counter aware.

Further advantage of this embodiment is that the distance counter only has local significance in the G.8032 ring, so that all other nodes attached to access ports of the ring are not aware of the distance counter.

The distance may be integrated in the header of the frame. In case the header of data frame does not have spare space for this purpose, it is proposed to integrate it into the pay load field of the frame, extending the size of the frame for example by one byte. The advantage of the integration in the payload field is that the non the G.8032 devices receiving the frame is solution is that may rely on a well defined frame header.

In a further embodiment it is proposed to add another VLAN tag into the frame header. The header of a G.8032 frame includes a VLAN tag used for traffic separation or prioritization. Thus, the data traffic may be assigned to different quality levels for traffic distribution wherein the level is recognised by the set VLAN tag. Adding another VLAN tag onto the tag stack may offer a possibility to carry additional data along the ring. The advantage of this solution is that the header is not exceeded and the existing fields in the header are used for other purpose while still easily being recognizable by the non G.8032 nodes due to its standardized structure.

In other embodiment it is proposed to alter fields of the frame header. Since the data traffic in the Ethernet ring remains in the Ethernet ring and an Ethernet ring may be seen as an administrative domain that does not require interaction to other authorities. This makes it possible to alter fields of the frame header, that are by definition fixed inside the ring and are not influencing the frame's per hop behavior.

In a further embodiment it is proposed to develop or extend a protocol, that is responsible to distribute the host distance counter HDC and the ring size counter over the ring. Preferably the distance counter may be integrated in an existing R-APS message. The distance counter may be integrated into the header of the R-APS message, for example there is the sub-code filed or the status reserved filed or reserved 2 filed as some example for integration.

While the current invention has been described in relation to its preferred embodi- ments, it is to be understood that this description is for illustrative purposes only. Accordingly, it is intended that the invention be limited only by the scope of the claims appended hereto.

Claims

1. A method for controlling data traffic distribution in an Ethernet ring node (100) being part of an Ethernet ring (10), the Ethernet ring node being configured to connect at least one host (130) and comprising two ring ports (110,120), each ring port adapted to connect an adjacent Ethernet ring node and the Ethernet ring node comprising an assignment entry (140) with an assignment between each ring port and a further host in the Ethernet ring being reachable over said ring port, the method comprising the steps of:

- providing distance information including information on a distance between the ring port and the host, wherein the distance information is included in the assignment entry (S32),

- detecting a topology change of the Ethernet ring by detecting a change point in the distribution of the data traffic in the Ethernet ring, wherein the detection is based on the reception of information providing the location of the change point (S33),

- determining any change in a reachability of the host over the ring port after the topology change wherein the determination is based on the distance information and the location of the change point (S35),

- based on the determination result (S36), performing an updating procedure on the assignment entry considering the location of the change point (S37) if the reachability of the host has changed.

2. Method according to claim 1 wherein the host is connected by means of a further port.

3. Method according to claim 1 wherein the distance information is determined from a host distance counter received on the ring port, wherein the host distance counter provides a number of passed Ethernet ring nodes between the ring port and the host.

4. Method according to claim 1 wherein the location of the change point is determined form a change point distance counter received on the ring port wherein the distance counter provides a number of passed Ethernet ring nodes between the change point and the ring port.

5. Method according to claim 1 wherein the determination step is based on the comparison of the change point distance counter and the host distance counter.

6. Method according to claim 1 wherein the updating procedure comprises changing the assignment between the ring port and the host.

7. Method according to claim 1 wherein the method comprises providing a size of the Ethernet ring by using a size distance counter received on the ring port.

8. Method according to claim 6 or 7 wherein the updating procedure comprises updating the distance information in the assignment entry by using the stored distance information and the ring size.

9. Method according to one of the claims 6 to 8 wherein the updating procedure comprises deleting the assignment of the host to the ring port.

10. Method according to any of the preceding claims wherein the host distance counter, the size distance counter and the change point distance counters are integrated in a data frame of the data traffic.

11. Method according to any of the preceding claims wherein the host distance counter and the size distance counter are integrated in a signalling message propagated in the Ethernet ring.

12. An Ethernet ring node (100) adapted to control data traffic distribution and being part of an Ethernet ring (10), the Ethernet ring node comprising:

- two ring ports (110,120), each ring port adapted to connect an adjacent Ethernet ring node,

- at least one further port being configured to connect at least one host (130),

- an assignment entry (140) being adapted to provide an assignment between each ring port and a further host in the Ethernet ring being reachable over said ring port and a distance information including information on a number of Ethernet nodes between the ring port and the host,

- a detector (150) adapted to detect a topology change of the Ethernet ring by detecting a change point in the data traffic in the Ethernet ring, wherein the detection is based on the reception of an information providing the location of the change point,

- a control unit (160) adapted to determine a reachability of the host over each ring port after the topology change wherein the examination is based on the distance information and the location of the change point,

- a processor (170) adapted to perform based on the examination result an updating procedure on the assignment entry considering the location of the change point if the reachability of the host has changed.