CN100369436C

CN100369436C - Looped network and its protecting method

Info

Publication number: CN100369436C
Application number: CNB200510132039XA
Authority: CN
Inventors: 宁军; 刘剑锋; 姚民; 郑涛
Original assignee: Hangzhou H3C Technologies Co Ltd
Current assignee: New H3C Information Technologies Co Ltd
Priority date: 2005-12-16
Filing date: 2005-12-16
Publication date: 2008-02-13
Anticipated expiration: 2025-12-16
Also published as: CN1791049A

Abstract

The present invention discloses a method for protecting a ring network comprising at least two nodes, wherein each node is connected to a ring through a first ring upper port and a second ring upper ring, and all the ring upper ports are mutually communicated in a link layer; the ring upper ports are communicated with at least one other port of all the nodes in a network layer. The method comprises the steps that the link layer provided with the ring upper port executes forwarding in a broadcasting mode; when the ring network generates a fault, two nodes which work normally and are contiguous to the fault point connect the ring of the upper port ring of the fault point back to another ring upper port. When the ring network generates the fault and the fault recovers, update of a forwarding table and fault notification are needed; a main node, a main port and auxiliary ports do not need to be arranged in the ring network, a blockage port is also needed; besides, the present invention achieves high-speed recovery and switching of below 50 ms.

Description

Looped network and guard method thereof

Technical field

The present invention relates to network communication field, relate in particular to a kind of link layer looped network guard method, use this method looped network and on the network equipment.

Background technology

Along with the development of ethernet technology in metropolitan area network and Access Network field, ethernet ring network is more and more used.Ethernet ring network is a kind of link layer network of redundancy structure, has failover and fault recovery function.Spanning-Tree Protocol (STP, Spanning Tree Protocol) can be used for realizing the failover and the recovery of looped network, but STP can only support 7 nodes, and the convergence rate of its switching and recovery reaches tens of seconds, and the service supplier that can not satisfy metropolitan area network provides the demand of carrier class fault-tolerant ability to the user.

Available technology adopting EAPS (Ethernet Automatic Protected Switching, Ethernet APS) realizes the protection of ethernet ring network.A node on the EAPS looped network is set to host node, the port blocking-up that host node will be connected with looped network when the looped network operate as normal; When looped network breaks down, the node adjacent with the fault point notified host node with ring network fault, host node is received the port of opening blocking-up behind the signalling trouble, and transmission refreshes FDB (Forwarding Database, transmit) message makes ring go up all node updates and transmits, and upgrade the back failover and finish transmitting.Fault recovery process similarly.

Adopt EAPS, when breaking down, fault detect, signalling trouble, refreshing of FDB all need node cpu (Central Process Unit, central processing unit) to handle, and message is transmitted and must just can be switched to after said process is finished on the normal link, otherwise business can not be recovered.As seen, EAPS failover speed dependent is in the response speed of node cpu, thereby the failover of EAPS is difficult to reach the fault-tolerant demand of carrier class of 50ms (millisecond) in the practical application.

Summary of the invention

The present invention will solve is need carry out signalling trouble and refresh the problem of transmitting during the ethernet ring network failover in the prior art.

The guard method of looped network among the present invention; described looped network is an ethernet ring network; described looped network comprises at least two nodes; each node is connected on the ring by the first ring upper port and the second ring upper port; all ring upper port are in the link layer intercommunication; the ring upper port is communicated with in network layer with at least one other port of all nodes, said method comprising the steps of:

The link layer forwarding that the ring upper port is set is carried out with broadcast mode;

When looped network broke down, the ring upper port that the node of two operate as normal adjacent with the fault point will be connected the fault point was looped back to another ring upper port.

Preferably, the first ring upper port of described each node of looped network is connected to the second ring upper port of an end adjacent node, and its second ring upper port is connected to the first ring upper port of other end adjacent node;

Described method also comprises:

The message that node is transmitted last ring is transmitted from the first ring upper port;

When node when the link layer address of source link layer address and this node is identical from the message that the second ring upper port receives, abandon this message.

Alternatively, described ring network fault is a node failure;

Described method also comprises: after two nodes that link to each other with former fault node are received the fault recovery notice message of former fault node transmission, connecting the ring upper port cancellation loopback of former fault node.

Alternatively, described ring network fault is a node failure;

Described method also comprises:

Former fault node starts the work of ring upper port after the working method configuration of its ring upper port is finished;

Two nodes that link to each other with former fault node are being connected the ring upper port cancellation loopback of former fault node.

Alternatively, described ring network fault is a link failure;

Described method also comprises: when ring network fault was all recovered two nodes adjacent with former fault point, described two adjacent nodes were connecting the ring upper port cancellation loopback of former fault point.

Preferably, described method also comprises:

On each node, safeguard the current topological database of looped network;

The link layer address of all nodes abandons this message not simultaneously in current topological database on source link layer address and this node in the message that node receives.

Alternatively, when described ring network fault, described method also comprises: node will be redirected to the second ring upper port from the broadcasting packet that the first ring upper port receives and send.

Preferably, described message is a multicast message;

Described method also comprises:

The first ring upper port of multicast node is added multicast group, at the second ring upper port startup network layer multicast of multicast node; Described multicast node is carried out the node that network layer is transmitted for the multicast message to this multicast group;

When described ring network fault, multicast node will be redirected to the second ring upper port from first multicast message that encircles this multicast group of upper port reception and send.

Preferably, described ring network fault is a link failure;

Described method also comprises:

Two nodes adjacent with former fault point are cancelled loopback in the ring upper port that is being connected with former fault point after confirming that former fault link all recovers both sides;

The node adjacent with former fault point abandons the message that receives from the second ring upper port after the cancellation loopback, before current topological database Central Plains faulty link recovers.

Preferably, describedly confirm that at two nodes former fault link all recovers back cancellation loopback to both sides and is specially:

The node that detects link-recovery sends link-recovery to another node adjacent with former fault point and seeks the opinion of message;

Receive node that link-recovery seeks the opinion of message when connected link also recovers, return link recovers confirmation message;

The node that receives the link-recovery confirmation message is cancelled loopback in the ring upper port that is connected with former fault point, and sends cancellation loopback notice message to another node adjacent with former fault point;

The node that receives cancellation loopback notice message is cancelled loopback in the ring upper port that is being connected with former fault point;

Described after the cancellation loopback, the faulty link node that abandons the message that receives from the second ring upper port before recovering in current topological database Central Plains is the node that receives the link-recovery confirmation message.

Preferably, described ring network fault is a node failure;

Described method also comprises:

After former fault node recovers, recover the notice message to its adjacent node sending node;

Receive node and recover the node of notice message at its ring upper port that is connected with this message source node cancellation loopback;

Former fault node with before being connected of two adjacent nodes recovers, abandons the message that receives from the second ring upper port after sending node recovers the notice message, in the current topological database.

Preferably, described former fault node recovers to be specially: former fault node starts the work of ring upper port after the working method configuration of its ring upper port is finished;

Described method also comprises: node recovers the notice message detecting and recovering back startup timing being connected of former fault node if receive node in the scheduled time, then at its ring upper port that is connected with former fault node cancellation loopback.

A kind of network equipment that is applied to looped network provided by the invention, comprise the first ring upper port and the second ring upper port that connect on the link layer looped network, described looped network is an ethernet ring network, it is characterized in that, also comprise fault detection unit, port loopback unit and agreement retransmission unit, wherein:

Fault detection unit is used for detecting the operating state of the part looped network that the first ring upper port and the second ring upper port be connected, and breaks down and notify port loopback unit when confirming fault recovery detecting;

Port loopback unit connection protocol retransmission unit is used for when receiving signalling trouble, and the ring upper port loopback that the part looped network that breaks down is connected encircles upper port to another; Receiving fault recovery when notice, the loopback cancellation of the ring upper port that the part looped network of fault recovery is connected;

The agreement retransmission unit is used for receiving protocol processes, the network layer of message and transmits and the link-layer broadcast forwarding.

Preferably, the forwarding of the network layer of described agreement retransmission unit is the inlet of looped network to unicast message with the first ring upper port;

The described network equipment also comprises the clean culture processing unit, connect second ring upper port and the agreement retransmission unit, be used for from the unicast message that the second ring upper port receives, identifying the source link layer address message identical, and abandon this message with the link layer address of this network equipment; Export other unicast messages to the agreement retransmission unit.

Preferably, the described second ring upper port is the member port of at least one multicast group; Described agreement retransmission unit carries out network layer to multicast message to be transmitted, and is outbound port to the multicast message that enters looped network with the second ring upper port;

The described network equipment also comprises the multicast processing unit, connect two ring upper port and agreement retransmission unit respectively, be used for identifying the source link layer address message identical the multicast message of its place multicast group of receiving from the first ring upper port, and abandon this message with the link layer address of this network equipment; To export the first ring upper port to from second multicast message that encircles its place group of upper port reception sends; Export other multicast messages to the agreement retransmission unit.

Preferably, the described network equipment also comprises the ring-network topology maintenance unit, connects the multicast processing unit, is used for setting up and safeguarding the current topology information of looped network;

The multicast processing unit identifies the different message of link layer address of all-network equipment on source link layer address and the current looped network the multicast message that receives from the first ring upper port, and abandons this message.

Alternatively, described ring-network topology maintenance unit connects described clean culture processing unit, the clean culture processing unit identifies the different message of link layer address of all devices on source link layer address and the current looped network the unicast message that receives from the second ring upper port, and abandons this message.

Preferably, described agreement retransmission unit is outlet to the broadcasting packet of looped network with the second ring upper port;

The described network equipment also comprises the broadcasting processing unit, connect two rings upper port, agreement retransmission unit and ring-network topology maintenance units respectively, be used for abandoning the source link layer address message identical from the broadcasting packet that the first ring upper port receives with the link layer address of this network equipment; To export the first ring upper port from the broadcasting packet that the second ring upper port receives to sends; Abandon from the broadcasting packet that the first ring upper port receives the different message of link layer address of all devices on the source link layer address and current looped network; Export other broadcasting packets to the agreement retransmission unit.

Preferably, the described network equipment also comprises the fault recovery processing unit, before this fault is eliminated after fault detection unit is confirmed fault recovery, in the topology information of ring-network topology maintenance unit, abandon multicast message that receives from the first ring upper port and the unicast message that receives from the second ring upper port.

A kind of message forwarding method that is applied to the link layer looped network node provided by the invention, described node comprises that node is connected the ring upper port of first on the looped network and second encircles upper port, and at least one ring external port, the node of connecting ring external port is communicated with in network layer with described looped network, may further comprise the steps:

To be forwarded to the message of looped network in the first ring upper port transmission in network layer;

The message that carries out the link layer forwarding is sent with broadcast mode;

When the part looped network that is connected with node broke down, the message loopback that the ring upper port of the part looped network that will be broken down by connection sends to another ring upper port sent.

Preferably, described method also comprises: abandon the source link layer address message identical with this node link layer address that the second ring upper port receives.

Preferably, described method also comprises: the different message of link layer address that abandons all nodes that can be communicated with on source link layer address that the second ring upper port receives and the current looped network.

Preferably, described message comprises unicast message;

Described method also comprises: if the purpose link layer address of the unicast message that receives is identical with this node link layer address, this message is carried out network layer transmit.

Preferably, described message comprises broadcasting packet;

Described method also comprises: to the broadcasting packet that receives from the first ring upper port, it is redirected to the second ring upper port sends.

Preferably, described message comprises multicast message;

Described method also comprises:

To the multicast message that receives from the second ring upper port, if the first ring upper port be the member port of this message place multicast group, then this message is carried out network layer and transmit or protocol processes, and it is copied to the first ring upper port transmission; Otherwise this message is carried out link-layer broadcast to be transmitted;

Multicast message to its place multicast group of receiving from the first ring upper port is redirected to the second ring upper port with it and sends.

Preferably, described method also comprises: after cancellation port loopback, confirm that part looped network that former loopback port is connected normally E-Packets before, abandon the message that receives from the second ring upper port.

A kind of link layer looped network provided by the invention, comprise at least two nodes, the first ring upper port of each node is connected to the second ring upper port of an end adjacent node, its second ring upper port is connected to the first ring upper port of other end adjacent node to form looped network, all nodes comprise that at least one ring external port connects the network that is communicated with in network layer with looped network on the ring, wherein:

Source node will send in the first ring upper port in the unicast message that network layer is forwarded to looped network, and described source node is the initial node that sends this message in looped network;

What node will receive from the second ring upper port is that the unicast message of source node abandons with it;

The link layer of each node is transmitted and is carried out with broadcast mode;

When the part looped network that connects breaks down, the node adjacent with the fault point will be looped back to another ring upper port by the message that the ring upper port that is connected the fault point sends and send.

Preferably, to multicast message:

Source node will be forwarded to the multicast message of looped network in the second ring upper port transmission in network layer;

When node receives the multicast message of its second ring upper port place multicast group from the first ring upper port,, abandon this multicast message if this multicast message is a source node with it; Otherwise this multicast message is carried out network layer forwarding or protocol processes and duplicates this multicast message sending from the second ring upper port;

When node receives the multicast message of this port place multicast group from the second ring upper port, this multicast message is redirected to the first ring upper port sends.

Preferably, the node different unicast message of all nodes that will can be communicated with on second source node that receive of ring upper port and the current looped network abandons;

The different multicast message of all nodes that node will can be communicated with on first source node that receive of ring upper port and the current looped network abandons.

Preferably, when ring network fault is recovered, the node that carries out the port loopback because of former fault point with ring upper port that former fault point is connected on cancel loopback.

The another kind of network equipment provided by the invention comprises two ring upper port that connect on looped network, and signalling trouble unit, loopback unit and protocol processing unit, and described looped network is an ethernet ring network, wherein:

The signalling trouble unit is used for detecting two operating states of encircling the part looped network of upper port connection, and sends signalling trouble when the part looped network that detects certain ring upper port connection breaks down;

The loopback unit is used for receiving signalling trouble, and the unicast message that will transmit from the ring upper port of above-mentioned connection looped network fault point originally is looped back to another ring upper port forwarding;

Preferably, the described network equipment also comprises the packet filtering unit, and what be used for that filtering receives for the second time is the unicast message of source node with the described network equipment.

The present invention adopts broadcast mode to carry out link layer on looped network and transmits, utilize the port loop fuction of looped network node simultaneously, should put another ring upper port forwarding of the message difference loopback of adjacent ring upper port transmission by link failure, and make message can arrive its destination node to its place node; The present invention does not need renewal to transmit when failover and carries out signalling trouble, and fault detect of the present invention simultaneously need not handled by node cpu, has realized being lower than quick switching and the recovery of 50ms;

Further, the present invention passes through message is transmitted from a ring upper port, is discarded in the message by the transmission of this node that another ring upper port receives simultaneously, can realize loop network link failover and recovery to all messages;

Further, the message that the present invention sends by the filter faults node can realize that node failure switches and recovery to all messages.

Description of drawings

Fig. 1 is the representative network topology example figure of looped network;

Fig. 2 is the flow chart of ring net protection method of the present invention;

Fig. 3 is the flow chart of loop network link fault recovery method of the present invention;

Fig. 4 is the flow chart of looped network node fault recovery method of the present invention;

Fig. 5 is the structure chart of the network equipment on the looped network of using guard method of the present invention;

Fig. 6 is the message forwarding process figure of the looped network node of application guard method of the present invention.

Embodiment

Looped network is connected in series one by one by the node of forming looped network and forms, and its typical networking structure can be referring to Fig. 1.Fig. 1 is the looped network of being made up of 4 nodes, and node 110,120,130 and 140 respectively has two ports to be connected on the looped network, is referred to as to encircle upper port; Each node also has at least 1 port to be used for connecting looped network network in addition usually, is referred to as to encircle external port, and promptly port one 13,123,133,134,143 and 144.

The applied looped network of the present invention is two layers of looped network, and promptly all ring upper port are in link layer connectivity.And be communicated with in network layer between ring network that external port connected and the looped network.Message is referred to as ring from ring forwarded to looped network that external port connected transmits, and message is forwarded to the ring network that external port connected from looped network and is referred to as down ring and transmits, going up then that ring transmits and encircle forwarding down all is three layers of forwarding.In this article, the node that the start of message (SOM) is sent in looped network is called the source node of this message, and initial the transmission in looped network comprises that message is transmitted, generated to ring and in being sent to looped network.

To the message that last ring is transmitted, its outlet may be any ring upper port of forward node, along the different destination nodes that arrive this message around direction.The ring upper port of each node is classified by its outlet message forwarding direction, and then port one 11,121,131 and 141 is a class, and its outlet message is transmitted along clockwise direction; Port one 12,122,132 and 142 is a class, and its outlet message is transmitted in the counterclockwise direction.Claim that in the present invention this two generic port is the first ring upper port and the second ring upper port.As seen, each node at one end holds the second ring upper port of adjacent node to connect by the first ring upper port therewith, is connected with the first ring upper port of those end adjacent nodes by the second ring upper port at the other end.

In embodiments of the invention one, two layers of looped network are used for transmitting unicast message.The flow process of ring net protection method is in the present embodiment:

The link-layer broadcast of carrying out that its ring upper port is set on each node is transmitted;

Looped network breaks down, and the link between certain two node connects interruption;

The ring upper port loopback that the node on faulty link both sides connects faulty link encircles upper port to another.

In the present invention, the port loopback on the node is the loopback of physical layer, and port A loopback is meaned that to port B the message that sends from port A is directly received by port A, sends out node from port B with it again.If node configuration the message that sends from port A and receive is carried out special processing, then this message will pass through same processing procedure before sending out port B.

The last unicast message of encircling forwarding is a destination node with node on the looped network beyond the source node, when destination node receives this unicast message, this message is carried out three layers of ring forwarding down.Therefore, when being used for transmitting unicast message on the looped network among the present invention, under the situation of operate as normal, can not form broadcast storm.Simultaneously, the source node of unicast message can encircle the outlet that upper port is a unicast message with any one, and unicast message arrives destination node along routing direction.

The node that receives unicast message is the link layer address of purpose link layer address and this node in the message relatively, if identical then this node be the destination node of this message, this message carried out three layers encircle forwarding down; If different, then carry out link-layer broadcast and transmit, message is sent to next node from another ring upper port.Link layer address can be the MAC Address of node three layer interfaces, also can be the three layer interface MAC Address and VLAN (Virtual Local Area Network, the VLAN) sign of node.

Node judges that message is carried out L 2 broadcast to be transmitted or three layers of forwarding, and carries out that L 2 broadcast is transmitted and process the present invention of three layers of forwarding is same as the prior art.The L 2 broadcast that the ring upper port is set is transmitted and can be realized by forbid the MAC address learning that encircles upper port on node; When all ring upper port belong to a VLAN, also can realize by the MAC address learning of on each node, forbidding this VLAN.

Easy for describing, see also the looped network among Fig. 1, when the link occurs fault between node 120 and the node 130, node 120 with port one 21 loopbacks to port one 22, node 130 with port one 32 loopbacks to port one 31.At this moment, node 140 is forwarded to port one 41 from port one 44 with certain unicast message ring, and the destination node of this message is 130.This message is forwarded to node 120 by node 110 from port one 11 broadcasting, and node 120 should send this message from port one 21, and to port one 22, this message is sent to node 110 from port one 22 at port one 21 loopbacks; This message is forwarded to node 130 by node 110 and node 140 broadcasting; 130 pairs of these messages of node carry out three layers of ring forwarding down.Like this, the forwarding of unicast message is not subjected to the influence of link failure.

In embodiments of the invention two, two layers of looped network are used for transmitting unicast message.The flow process of ring net protection method is in the present embodiment:

Its ring upper port is set on each node carries out the link-layer broadcast forwarding;

Node will be gone up the message of ring forwarding and transmit from the first ring upper port;

What node will receive from the second ring upper port is the packet loss of source node with this node, and the source link layer address of this message is identical with the link layer address of this node;

Looped network breaks down;

The ring upper port loopback that the node on both sides, fault point connects the fault point encircles upper port to another.

In the present embodiment, the ring forwarding of going up of all messages is all undertaken by the direction of setting.Abandon certain the class message that receives from the second ring upper port and can issue the realization of corresponding ACL (AccessControl List, Access Control List (ACL)) rule by encircling upper port second.

With looped network shown in Figure 1 is example, and when node 120 broke down, the node adjacent with the fault point was node 110 and node 130, node 110 with port one 11 loopbacks to port one 12, node 130 with port one 32 loopbacks to port one 31.If port one 41 is the first ring upper port, node 140 will be that the unicast message ring of destination node is forwarded to port one 41 with node 120 from port one 44, since node 120 faults, the second ring upper port 142 that this message will extremely be connected with node 130 from port one 41 loopbacks that are connected with node 110; Node 140 receives the message of source node for this node from the second ring upper port 142, promptly abandons this message, and no longer from port one 41 broadcasting, to avoid this message repetitive cycling on looped network.Like this, the forwarding of looped network unicast message still can operate as normal under the situation of node failure and link failure in the present embodiment.

Node can adopt 802.3ah or UDLD (UnidirectionalLink Detection to the detection of ring network fault and recovery, one way link detects) etc. agreement carry out, break down or during fault recovery at the part looped network that is connected with node, the ring upper port that connects this part looped network detects to connect and lost efficacy or recovery, can notify this node by interrupt mode.

After ring network fault is recovered, carry out two node cancellation loopbacks of port loopback on the looped network because of this fault.When node failure in the present embodiment recovers; the ring upper port one of malfunctioning node starts; two nodes that are attached thereto will produce interruption; but in some cases; this moment malfunctioning node the ring upper port also the unfinished work mode dispose, the ring upper port only has basic forwarding capability and does not have and finish the required function of looped network protection flow process.If the loopback of adjacent node cancellation at this moment may cause unusual that unicast message is transmitted on the looped network.Can adopt following two kinds of methods to address this problem:

The first makes malfunctioning node send fault recovery notice message in the working method configuration back of finishing the ring upper port to adjacent node, and adjacent node is cancelled the port loopback again after receiving fault recovery notice message;

It two is boot sequences of revising node, finishes the working method configuration of port earlier, port is started again, and can guarantee like this to produce when interrupting on adjacent node, and former fault node can normally E-Packet.

During link failure recovery in embodiment one and present embodiment, because node is all in running order, processing to unicast message is normally carried out, can not form broadcast storm, thereby the ring upper port that is connected with former fault point recovers to connect back generation interruption on node, node can return to its normal operating conditions directly at this port cancellation loopback.

In embodiment three, two layers of looped network are used for transmitting unicast message and broadcasting packet.Increase following steps on the flow process in embodiment two:

The current topological database of looped network is set up and safeguarded to each node on looped network, keeps each nodal information that can be communicated with and neighbouring relations each other on the current looped network;

The source node that node judge to receive message whether in the topological database of current looped network, if not abandon this message;

To the broadcasting packet that receives in the first ring upper port, node is redirected to the second ring upper port with it and sends.

The current topological database of setting up looped network on node can adopt LLDP (Link LayerDiscovery Protocol, Link Layer Discovery Protocol) or HDP (Huawei Discovery Protocol, Huawei's Topology Discovery agreement) agreement to realize; Node can judge whether the source node that receives message is online by the source link layer address and the link layer address in the current ring-network topology database node information that relatively receive in the message; Node issues the acl rule realization to redirected also can the employing in the first ring upper port that the first ring upper port receives message.

When the looped network operate as normal, the broadcasting packet that each node sends can be along the first ring upper port behind other nodes of traversal, return the second ring upper port of source node around direction, since on each node all to being that the message of source node filters with this node, broadcasting packet can be on looped network repetitive cycling.

In the looped network of Fig. 1, when the link occurs fault between node 120 and the node 130, node 120 with port one 21 loopbacks to port one 22, node 130 with port one 32 loopbacks to port one 31.If port one 41 is the first ring upper port, node 140 is sent to node 110 with certain broadcasting packet from the first ring upper port 141, and is sent to node 120 by node 110, and

node

110 and 120 receives and handle this broadcasting packet from the second ring upper port in this process; When node 120 these messages of broadcasting, because port one 21 loopbacks, broadcasting packet is sent to the port one 11 of node 110 by port one 22 by node 120, this moment, the receiving port of this broadcasting packet became the first ring upper port, thereby this broadcasting packet directly is redirected to the second ring upper port and sent, and avoided the aftertreatment to this message; Send until port one 32 loopbacks to the port one 31 of this broadcasting packet from node 130, the second ring upper port 142 that arrives its source node 140 is dropped.Like this, can guarantee that broadcasting packet travels through each node and can not form broadcast storm simultaneously.

When node broke down, if malfunctioning node is not the source node of broadcasting packet, then two of this broadcasting packet layers of forwarding convergence process were consistent with above-mentioned link failure.If source node breaks down after sending broadcasting packet, delete source node in the topological database that the node of other operate as normal can be safeguarded thereon, and because of the source node of this broadcasting packet not in topological database, this broadcasting packet will be dropped as the message in the unknown source.

Looped network among the embodiment four is used for forwarding multicasting message, and the flow process of its guard method as shown in Figure 2.At step S210, the link layer forwarding that its ring upper port is set on each node is carried out with broadcast mode.

At step S220, the current topological database of looped network is set up and safeguarded to each node on looped network, keeps the information of each node that can be communicated with on the current looped network and neighbouring relations each other.

At step S230, to need the multicast data message to certain multicast group to carry out three layers of forwarding, its multicast protocol message is carried out the multicast node of the node of protocol processes for this multicast group on the looped network, the first ring upper port of multicast node is added this multicast group, at the second ring upper port startup network layer multicast of multicast node.

At step S240, node will be gone up the message of ring forwarding and transmit from the first ring upper port.

At step S250, the source link layer address packet loss identical with the link layer address of this node to receiving from the second ring upper port promptly sent or gone up and encircle the message of transmitting by this node, and node abandons it.

At step S260, looped network breaks down.

At step S270, the ring upper port loopback that the node on the both sides, fault point connects the fault point encircles upper port to another.

At step S280, multicast node is redirected to the second ring upper port with it and sends the multicast message of its place multicast group of receiving in the first ring upper port;

At step S290, to the source node that the receives multicast message in the topological database of current looped network not, node is with the packet loss of this message as the unknown source.

Be example still, establish the multicast message of a multicast group and will transmit that transmit from

node

120 and 130 times rings of node, then node 120,130 and 140 is that the multicast node and the node 140 of this multicast group are upstream node from node 140 rings with looped network shown in Figure 1.If port one 41 is the first ring upper port of node 140, then port one 21 and port one 31 are respectively the first ring upper port of node 120 and 130.Port one 21,131 and 141 is added this multicast group, become the member port of this multicast group, correspondingly, on port one 22,132 and 142, start three layers of multicast.

Under the situation of looped network operate as normal, node 140 will be transmitted from ring on the multicast data message of ring external port, send on the looped network from port one 41; The port of node 110 is not the member of this multicast group, this multicast data message is carried out L 2 broadcast transmit, and is sent to the port one 22 of node 120 from port one 11; Owing to start network layer multicast on port one 22, port one 21 still be the member port of this multicast group simultaneously, and the ring forwarding down of 120 pairs of these multicast data messages of node will be duplicated the line link layer forwarding of going forward side by side of this multicast data message to port one 21 simultaneously; The processing of 130 pairs of these messages of node is identical with node 120; When node 140 in the second ring upper port 142 when receiving this multicast data message that traveled through looped network node, because of this message is that source node abandons it with this node.

When looped network generation link failure or node failure, to the processing of multicast data message except to carry out in multicast node three layers transmit and duplicate, among all the other processes and the embodiment three processing of broadcasting packet be there is no different, no longer repetition herein.

Looped network when operate as normal, link failure or node failure to the multicast protocol message, except on multicast node, carrying out three layers of forwarding to multicast data message, then should carry out correspondingly outside the protocol processes the multicast protocol message according to its agreement, identical with processing procedure to multicast data message, no longer repeat.

Necessary each the online node of traversal ring of multicast message and broadcasting packet, and each node should only carry out an agreement or three layers of forwarding processing to multicast message and broadcasting packet.Arrive destination node with unicast message and promptly transmitted and compare by ring down, more more complicated to the processing of multicast message and broadcasting packet, especially to note avoiding the broadcast storm that may cause.

Therefore, when the fault recovery of embodiment three and embodiment four, can adopt different recovery flow processs according to the difference of fault point.

When link failure was recovered, faulty link might at one end recover, and did not recover as yet at the other end, connected the interruption cancellation port loopback that recovery produces if recover the node of an end this moment according to the ring upper port, can cause the abnormality of looped network work.So two nodes adjacent with former fault point are just cancelled loopback in the ring upper port that is being connected with former fault point after confirming that former fault link has all recovered to both sides.In addition, because two nodes might not be to cancel the port loopback simultaneously, the looped network operation irregularity that may cause in this very of short duration process, make the node adjacent after the cancellation loopback, before current topological database Central Plains faulty link recovers, abandon the message that receives from the second ring upper port with former fault point.

Figure 3 shows that the recovery flow process of link failure, wherein first adjacent node is the adjacent node that detects link failure recovery, and second adjacent node is another adjacent node of this faulty link.At step S310, first adjacent node detects faulty link and recovers.

At step S320, first adjacent node sends link-recovery to second adjacent node and seeks the opinion of message, and whether inquiry the other side's link recovers when notifying the second adjacent node one's own side link-recovery.

At step S330, if second adjacent node has detected connected link-recovery, then recover confirmation message to the first adjacent node return link, inform that the other side one's own side's link also recovers; Otherwise the link-recovery that receives not being seeked the opinion of message replys.

At step S340, after first adjacent node receives the link-recovery confirmation message,, send cancellation loopback notice message to second adjacent node simultaneously connecting the ring upper port cancellation loopback of former fault link, instruct second adjacent node also to cancel the port loopback.

At step S350, first adjacent node after the cancellation loopback, this node is before topological database Central Plains faulty link recovers before being taken in, and abandons the message that receives from the second ring upper port.

At step S360, after second adjacent node receives cancellation loopback notice message, connecting the ring upper port cancellation loopback of former fault link.

If first adjacent node does not receive the link-recovery of second adjacent node and seeks the opinion of message, then do not cancel the port loopback.As long as the state continuance of link-recovery, first adjacent node will repeat to send link-recovery with the predetermined cycle of seeking the opinion of and seek the opinion of message, up to the link-recovery confirmation message of receiving second adjacent node.

For fear of link failure when just recovering because of the unstable connected sum that occurs repeatedly connects situation about disconnecting, can make adjacent node when detecting link-recovery, wait for certain hour after, send link-recovery again and seek the opinion of message and start fault recovery process.

Figure 4 shows that the recovery flow process of node failure.At step S410, former fault node recovers.Former fault node can start port after the working method configuration of finishing port again.

At step S420, former fault node recovers the notice message to two ends adjacent node sending node, informs that the fault of two ends this node of adjacent node is recovered.

At step S430, former fault node after sending node recovers the notice message, this node be taken in before in the topological database with before being connected of two adjacent nodes recovers, abandon from the message of the second ring upper port reception.

At step S440, the adjacent node of former fault node receives node and recovers the notice message.

At step S450, receive node and recover to notify the node of message to cancel loopback in the ring upper port that is being connected with this message source node.

Equally, the situation for fear of node failure breaks down repeatedly and recovers because of instability when just recovering can make former fault node wait for certain hour after recovery, and sending node recovers the process that the notice message starts the node recovery again.

In addition, dispose the node that the notice of sending node recovery afterwards message is finished in startup with port, can adopt timeout mechanism to open loopback, make former fault node finish startup and layoutprocedure so that reserve time enough to not being supported in the port working mode.Since when former fault node recovers with foundations that is connected of adjacent node, can generation interruption on adjacent node.Can make adjacent node in the startup timing of having no progeny, recover the notice message if in a setting-up time, receive the node of former fault node, then cancel loopback at the port that is being connected with former fault node.

Figure 5 shows that the structure chart of the network equipment of working on the looped network of using above-mentioned guard method, fault detection unit 503, port loopback unit 504, agreement retransmission unit 505, ring-network topology maintenance unit 509, fault recovery processing unit 510 are connected between the first ring upper port 501 and the second ring upper port 502; Port loopback unit 504 connects fault detection unit 503 and agreement retransmission unit 505 respectively; Clean culture processing unit 506, multicast processing unit 507, broadcasting processing unit 508 are connected between agreement retransmission unit 505, ring-network topology maintenance unit 509 and the fault recovery processing unit 510.

The operating state of the part looped network that the first ring upper port 501 of 503 pairs of network equipments of fault detection unit is connected with the second ring upper port 502 detects.When the part looped network that connects when detecting broke down, the ring upper port notice port loopback unit 504 that is connected the fault point took place fault and in fault detection unit 503; The ring upper port loopback that port loopback unit 504 will connect the fault point encircles upper port to another, should will encircle upper port from another by the message that agreement retransmission unit 505 sends from the ring upper port that connects the fault point and send.When the part looped network of confirming to break down recovered operate as normal, fault detection unit 503 started 510 work of fault recovery processing unit simultaneously with fault recovery and the ring upper port notice port loopback unit 504 that is connected former fault point; Loopback is cancelled in port loopback unit 504 on the ring upper port that connects former fault point, message is normally sent by agreement retransmission unit 505.Fault detection unit 503 can come according to the fault recovery method among embodiment two or embodiment three and the embodiment four whether the part looped network that breaks down is recovered operate as normal and confirm.

Agreement retransmission unit 505 is broadcasted forwarding at link layer to message, protocol massages is correspondingly handled according to the agreement regulation, and carried out network layer and transmit.

505 pairs of messages that ring is transmitted on network layer of agreement retransmission unit can be that outbound port enters looped network with the first ring upper port 501, can be that outbound port enters looped network with the second ring upper port 502 also.Be dispense flow rate suitably, can be according to being forwarded on the looped network from the first ring upper port 501 and the second ring upper port 502 respectively at the type of message that transmits on the looped network.For example, the outbound port that makes unicast message transmit for last ring with the first ring upper port 501, the outbound port that makes multicast message transmit for last ring with the second ring upper port 502 makes broadcasting packet send from the second ring upper port 502.

The current topological database of looped network is set up and safeguarded to ring-network topology maintenance unit 509, stores the information of each network equipment that can be communicated with on the current looped network and neighbouring relations each other, comprises the link layer address of the network equipment in institute's canned data.

To message from the first ring

upper port

501 and 502 receptions of the second ring upper port, earlier through fault recovery processing unit processes 510, again according to the type of message respectively after clean culture processing unit 506, multicast processing unit 507 or broadcasting processing unit 508 are handled, just can enter agreement retransmission unit 505.

Whether fault recovery processing unit 510 is being received the fault recovery notice back startup work of fault detection unit 503, constantly the former fault point that institute notifies recovery in the topology information of detection ring net topology maintenance unit 509 storages operate as normal.Before detecting this fault and eliminating in current topological database, the outbound port that fault recovery processing unit 510 is transmitted by agreement retransmission unit 505 according to the type that receives message and the type message is handled message.For example, when agreement retransmission unit 505 was going up ring and transferring port of unicast message with the first ring upper port 501, fault recovery processing unit 510 abandoned the unicast message that receives from the second ring upper port 502; When agreement retransmission unit 505 be going up ring and transferring port of multicast message and broadcasting packet with the second ring upper port 502, fault recovery processing unit 510 abandoned from the multicast message and the broadcasting packet of 501 receptions of the first ring upper port.In case fault recovery processing unit 510 detects the former fault point of recovering of notifying and eliminates in current topological database, promptly quit work, encircle the message of upper port 502 receptions freely by fault recovery processing unit 510 from the first ring upper port 501 and second this moment.

When agreement retransmission unit 505 is going up ring and transferring port of unicast message with the first ring upper port 501, the processing of 506 pairs of unicast messages of clean culture processing unit comprises: identifying with this node the unicast message that receives from the second ring upper port 502 is the unicast message of source node, abandons this message; When being connected with ring-network topology maintenance unit 509, clean culture processing unit 506 can identify the not message in current ring-network topology database of source node from the unicast message of the first ring

upper port

501 or 502 receptions of the second ring upper port, and abandons this message; To other unicast message, clean culture processing unit 506 exports it to agreement retransmission unit 505.

When agreement retransmission unit 505 with the second ring upper port 502 during for the going up ring and transfer port of the multicast message of certain multicast group, the second ring upper port 502 is member ports of this multicast group.The multicast message of the second ring upper port, the 502 place multicast group of 507 pairs of receptions of multicast processing unit is handled: if this multicast message receives from the first ring upper port 501, then abandoning wherein is the message of source node with present networks equipment; If this multicast message receives from the second ring upper port 502, then it is directly exported to the first ring upper port 501 and send; If multicast processing unit 507 is connected with ring-network topology maintenance unit 509, then will the not packet loss in current ring-network topology database of source node from the multicast message that the first ring upper port 501 receives; Other multicast messages export the agreement retransmission unit to by multicast processing unit 507.

When agreement retransmission unit 505 is the outbound port of broadcasting packet with the second ring upper port 502, the processing of 508 pairs of broadcasting packets of broadcasting processing unit comprises: identifying with present networks equipment the broadcasting packet that receives from the first ring upper port 501 is the message of source node, and it is abandoned; To directly export the first ring upper port 501 from the broadcasting packet that the second ring upper port 502 receives to sends; The broadcasting packet that receives from the first ring upper port 501, identify the not message in current ring-network topology database of source node, it is abandoned; Export other broadcasting packets to agreement retransmission unit 505.

Corresponding to the ring net protection method among aforementioned four embodiment; be understood that; when looped network was only protected the link failure of unicast message forwarding, the network equipment on the looped network only needed the first ring upper port 501, the second ring upper port 502, fault detection unit 503, port loopback unit 504 and agreement retransmission unit 505 just can realize its function.When looped network need be protected the node failure that unicast message is transmitted, increase clean culture processing unit 506 again and get final product.Corresponding to the ring network fault of type that E-Packets on the looped network and protection, can enumerate no longer one by one with reference to aforementioned each embodiment increase unit correspondingly herein.

To using the looped network of guard method of the present invention; node has the first ring upper port, the second ring upper port and at least 1 ring external port usually on it; the first ring upper port and the second ring upper port are connected this node on the looped network, are communicated with in network layer between ring network that external port connected and the looped network.Such node adopts flow process as shown in Figure 6 that message is transmitted on the looped network:

At step S601, node receives message.

At step S602, judge whether this message derives from the ring external port and transmit to looped network, if, execution in step S603; If not, execution in step S604.

At step S603, be that outbound port is gone up the ring forwarding to message with the first ring upper port, change step S615.

At step S604, judge that whether this message is after cancellation port loopback, confirms the message that part looped network that former loopback port is connected receives from the second ring upper port before normally E-Packeting, if abandon this message; If not, execution in step S605.

At step S605, judge that whether this message receives and be source node with this node from the second ring upper port, whether with the link layer address of this node identical, if abandon this message from the second ring upper port if promptly receiving the source link layer address of message; If not, execution in step S606.

At step S606, judge whether the source node of this message is the node that can be communicated with on the current looped network, the link layer address of one of node that can be communicated with on the source link layer address of promptly judging this message and the current looped network is identical, if not, abandons this message; If then continue.

Step S601 to the message among the step S606 can be a kind of or its combination in unicast message, broadcasting packet, the multicast message.

At step S607, to unicast message, judge whether the purpose link layer address of this message is identical with the link layer address of this node, if, execution in step S608; If not, execution in step S609.

At step S608, this unicast message is carried out ring forwarding under the network layer.

At step S609, to broadcasting packet, judge whether this message is received by the first ring upper port, if, execution in step S610; If not, execution in step S611.

At step S610, this message redirecting to the second ring upper port is transmitted, change step S615.

At step S611, to multicast message, judge whether this message is that first multicast message and this message that encircles upper port place multicast group received by the first ring upper port, if, execution in step S610; If not, change step S612.

At step S612, to multicast message, judge whether this message is that first multicast message and this message that encircles upper port place multicast group received by the second ring upper port, if, execution in step S613; If not, execution in step S614.

At step S613, this multicast message is carried out ring forwarding under the network layer, simultaneously this multicast message is duplicated the back and carry out the link-layer broadcast forwarding by the first ring upper port, change step S615.

At step S614, this message is carried out link-layer broadcast transmit.

At step S615, judge whether the part looped network that this node connects breaks down, if, execution in step S616; If not, this message is normally sent.

At step S616, the ring upper port loopback that connects the looped network fault point is sent this message to another ring upper port.

Above-described node message forwarding process is when this looped network is used for transmitting clean culture, multicast and broadcasting packet, can be used to protect the entire flow of link and node failure.Need to prove that each step S605, S606, S607, S609, S611, the S612 that in the flow process message characteristic are judged can sort arbitrarily,, can realize the present invention as long as the pairing message processing mode of its judged result is constant.

When looped network did not transmit the message of certain type, the step that relates to the type message can be omitted, and when the fault type of the required protection of looped network reduced, corresponding step also can be omitted.Omissible step those skilled in the art can reach a conclusion apparently with reference to aforementioned four embodiment, repeat no more herein.

The looped network of using guard method among the present invention is the link layer looped network, promptly at each node two layers of connection.Plural node is arranged on the looped network, and the first ring upper port of each node connects the second ring upper port of the node adjacent with this ring upper port, and the second ring upper port of each node connects the first ring upper port of the node adjacent with this ring upper port.Looped network is communicated with in network layer with other networks by at least one port of node on it.

The link layer of each node is transmitted and is carried out with broadcast mode on the looped network.When looped network broke down, the ring upper port loopback that two nodes adjacent with the fault point will be connected the fault point encircled upper port to another.When ring network fault is recovered, two nodes that carry out the port loopback because of this fault will connect the loopback cancellation of the ring upper port of former fault point.

To unicast message, source node is that outbound port is gone up the ring forwarding with the first ring upper port.Each node receives when being the unicast message of source node with this node from the second ring upper port, and this unicast message is abandoned; When receiving the purpose link layer address unicast message identical with this node link layer address, to this message carry out three layers down ring transmit; When receiving the different unicast message of all nodes that can be communicated with on source node and the current looped network from the second ring upper port, it is abandoned.

To multicast message, source node is that outbound port is gone up the ring forwarding with the second ring upper port.When each node receives the multicast message of the second ring upper port place multicast group of this node from the first ring upper port,, abandon this message if this multicast message is a source node with this node; Transmit or protocol processes otherwise this message is carried out three layers, and duplicate this message and carry out the link-layer broadcast forwarding from the second ring upper port; When receiving the multicast message of its place multicast group, this multicast message being redirected to the first ring upper port sending from the second ring upper port; When receiving the different multicast message of all nodes that can be communicated with on source node and the current looped network from the first ring upper port, it is abandoned.

To broadcasting packet, source node is that outbound port sends with the second ring upper port.Each node receives when being the broadcasting packet of source node with this node from the first ring upper port, abandons this message; When receiving broadcasting packet, this broadcasting packet being redirected to the first ring upper port sending from the second ring upper port; When receiving the different broadcasting packet of all nodes that can be communicated with on source node and the current looped network from the first ring upper port, it is abandoned.

To being used for transmitting the looped network of unicast message, the network equipment that is connected on the looped network by the ring upper port comprises interconnected signalling trouble unit and loopback unit.The signalling trouble unit detects the ruuning situations of the part looped network that two ring upper port connect, and when detecting certain ring part looped network that upper port connected and break down, sends signalling trouble to the loopback unit.After the loopback unit received signalling trouble, the unicast message loopback that the ring upper port of the part looped network that will be broken down by connection sends to another ring upper port sent.Like this, the link failure of looped network can be recovered.

For recovering the node failure of looped network, can in the above-mentioned network equipment, increase the packet filtering unit, connecting ring receipt unit, is that the unicast message of source node filters out with reception for the second time with present networks equipment.Receive for the second time when being the unicast message of source node with present networks equipment at the network equipment, illustrate that the destination node of this message breaks down, should give filtering this moment with this message.

The present invention transmits by port loopback and link-layer broadcast and realizes the fault-tolerant of looped network, is guaranteeing that the message of when each node only carries out single treatment to message message and the unknown of traversal looped network being originated filters, and has avoided broadcast storm.The present invention can realize the carrier class convergence rate of 50ms, and need not to be provided with host node, master port and from port, also need not port blocked in looped network; The present invention does not have specific (special) requirements to the port of node, can support various types of ports; On looped network, use the operation that the present invention can not influence other agreements simultaneously.

Above-described embodiment of the present invention does not constitute the qualification to protection range of the present invention.Any any modification of being done within the spirit and principles in the present invention, be equal to and replace and improvement etc., all should be included within the claim protection range of the present invention.

Claims

1. the guard method of a looped network; described looped network is an ethernet ring network; described looped network comprises at least two nodes; each node is connected on the ring by the first ring upper port and the second ring upper port; all ring upper port are in the link layer intercommunication; the ring upper port is communicated with in network layer with at least one other port of all nodes, it is characterized in that, said method comprising the steps of:

2. the guard method of looped network according to claim 1 is characterized in that: the first ring upper port of described each node of looped network is connected to the second ring upper port of an end adjacent node, and its second ring upper port is connected to the first ring upper port of other end adjacent node;

Described method also comprises:

3. as the guard method of looped network as described in the claim 2, it is characterized in that described ring network fault is a node failure;

4. as the guard method of looped network as described in the claim 2, it is characterized in that: described ring network fault is a node failure;

Described method also comprises:

5. the guard method of looped network as claimed in claim 1 or 2 is characterized in that: described ring network fault is a link failure;

6. as the guard method of looped network as described in the claim 2, it is characterized in that described method also comprises:

On each node, safeguard the current topological database of looped network;

7. as the guard method of looped network as described in the claim 6, it is characterized in that when looped network broke down, described method also comprised: node will be redirected to the second ring upper port from the broadcasting packet that the first ring upper port receives and send.

8. as the guard method of looped network as described in the claim 6, it is characterized in that: described message is a multicast message;

Described method also comprises:

When looped network breaks down, multicast node will be redirected to the second ring upper port from first multicast message that encircles this multicast group of upper port reception and send.

9. as the guard method of looped network as described in any one of the claim 6 to 8, it is characterized in that: described ring network fault is a link failure;

Described method also comprises:

10. as the guard method of looped network as described in the claim 9, it is characterized in that, describedly confirm that at two nodes former fault link all recovers back cancellation loopback to both sides and is specially:

11. as the guard method of looped network as described in any one of the claim 6 to 8, it is characterized in that: described ring network fault is a node failure;

Described method also comprises:

12. the guard method as looped network as described in the claim 11 is characterized in that, described former fault node recovers to be specially: former fault node starts the work of ring upper port after working method configuration of its ring upper port is finished;

13. a network equipment comprises the first ring upper port and the second ring upper port that connect on the link layer looped network, described looped network is an ethernet ring network, it is characterized in that, also comprises fault detection unit, port loopback unit and agreement retransmission unit, wherein:

14. the network equipment as claimed in claim 13 is characterized in that: it is the inlet of looped network with the first ring upper port that the network layer of described agreement retransmission unit is transmitted unicast message;

15. as the claim 13 or the 14 described network equipments, it is characterized in that: the described second ring upper port is the member port of at least one multicast group; Described agreement retransmission unit carries out network layer to multicast message to be transmitted, and is outbound port to the multicast message that enters looped network with the second ring upper port;

16. the network equipment as claimed in claim 15 is characterized in that: the described network equipment also comprises the ring-network topology maintenance unit, connects the multicast processing unit, is used for setting up and safeguarding the current topology information of looped network;

17. the network equipment as claimed in claim 16, it is characterized in that: described ring-network topology maintenance unit connects described clean culture processing unit, the clean culture processing unit identifies the different message of link layer address of all devices on source link layer address and the current looped network the unicast message that receives from the second ring upper port, and abandons this message.

18. the network equipment as claimed in claim 16 is characterized in that: described agreement retransmission unit is outlet to the broadcasting packet of looped network with the second ring upper port;

19. the network equipment as claimed in claim 16, it is characterized in that: the described network equipment also comprises the fault recovery processing unit, before this fault is eliminated after fault detection unit is confirmed fault recovery, in the topology information of ring-network topology maintenance unit, abandon multicast message that receives from the first ring upper port and the unicast message that receives from the second ring upper port.

20. message forwarding method that is applied to the link layer looped network node, described node comprises that node is connected the ring upper port of first on the looped network and second encircles upper port, and at least one ring external port, the node of connecting ring external port is communicated with in network layer with described looped network, it is characterized in that, may further comprise the steps:

21., it is characterized in that described method also comprises: abandon the source link layer address message identical that the second ring upper port receives with this node link layer address as being applied to the message forwarding method of link layer looped network node as described in the claim 20.

22. as being applied to the message forwarding method of link layer looped network node as described in the claim 21, it is characterized in that described method also comprises: the different message of link layer address that abandons all nodes that can be communicated with on source link layer address that the second ring upper port receives and the current looped network.

23., it is characterized in that described message comprises unicast message as being applied to the message forwarding method of link layer looped network node as described in the claim 22;

24. as being applied to the message forwarding method of link layer looped network node as described in the claim 22, it is characterized in that: described message comprises broadcasting packet;

25. as being applied to the message forwarding method of link layer looped network node as described in any one of the claim 22 to 24, it is characterized in that: described message comprises multicast message;

Described method also comprises:

26. as being applied to the message forwarding method of link layer looped network node as described in the claim 25, it is characterized in that, described method also comprises: after cancellation port loopback, confirm that part looped network that former loopback port is connected normally E-Packets before, abandon the message that receives from the second ring upper port.

27. link layer looped network, comprise at least two nodes, the first ring upper port of each node is connected to the second ring upper port of an end adjacent node, its second ring upper port is connected to the first ring upper port of other end adjacent node to form looped network, on the ring all nodes comprise at least one ring external port connect with looped network at the network that network layer is communicated with, it is characterized in that:

28. link layer looped network as claimed in claim 27 is characterized in that, to multicast message:

29. link layer looped network as claimed in claim 28 is characterized in that:

The different unicast message of all nodes that node will can be communicated with on second source node that receive of ring upper port and the current looped network abandons;

30., it is characterized in that as the described link layer looped network of claim 27 to 29: when ring network fault is recovered, the node that carries out the port loopback because of former fault point with ring upper port that former fault point is connected on cancel loopback.

31. a network equipment comprises two ring upper port that connect on looped network, described looped network is an ethernet ring network, it is characterized in that, also comprises signalling trouble unit, loopback unit and agreement retransmission unit, wherein:

32. the network equipment as claimed in claim 31 is characterized in that, the described network equipment also comprises the packet filtering unit, and what be used for that filtering receives for the second time is the unicast message of source node with the described network equipment.