CN101888340B

CN101888340B - Method for refreshing regional addresses of subring

Info

Publication number: CN101888340B
Application number: CN2010102304732A
Authority: CN
Inventors: 王斌
Original assignee: Nanjing Post and Telecommunication University
Current assignee: Nanjing Post and Telecommunication University; Nanjing University of Posts and Telecommunications
Priority date: 2010-07-19
Filing date: 2010-07-19
Publication date: 2012-11-28
Anticipated expiration: 2030-07-19
Also published as: CN101888340A

Abstract

The invention discloses a method for refreshing regional addresses of a subring, which comprises the following steps that: the subring divides a node region into two types of nodes, namely first-type nodes and second-type nodes; when the first-type nodes receive address refreshing messages sent by the first-type nodes or the second-type nodes receive address refreshing messages sent by the second-type nodes, the nodes refresh address forwarding tables which receive the address refreshing messages and are associated with a receiving port; and when the first-type nodes receive address refreshing messages sent by the second-type nodes or the second-type nodes receive address refreshing messages sent by the first-type nodes, the nodes refresh the address forwarding items which do not receive the address refreshing messages and are associated with the other subring port. The method reduces the range of addresses refreshing so as to reduce the hazards of data broadcasting storm and ensure that Ethernet multi-loop networks can converge within 50ms after links are reversed, and has important significance for improving the performance of the subring.

Description

A kind of regional address refreshing method of subring

Technical field

The present invention relates to data communication field, relate more specifically to a kind of regional address refreshing method of subring.

Background technology

Along with the development of Ethernet towards the multiple service supporting direction, particularly some business are increasingly high to reliability, the real-time requirement of network, Ethernet has extensively adopted annular networking to improve network reliability.And in the guard method of annular, require quick protective switch usually, reach below the 50ms.The technology of present this quick protective switch has IETF (Internet Engineering Task Force; The Internet engineering duty group) G.8032 RFC3619, ITU-T (InternationalTelecommunication Union, International Telecommunications Union) wait.

Definition for subring; The standard of formulating in the world (like ITU G.8032) thinks that subring (Sub-Ring) is a kind of ether ring that links to each other with other ring or network through interconnecting nodes (Interconnection Node), and interconnecting nodes (Interconnection Node) is the common node that belongs to two or more ether rings simultaneously.

For example shown in Figure 1, the node A to G on the subring Sub-ring1 is the node with Ethernet function of exchange, and Sub-ring1 is through interconnection contact access network X.User M is connected with Node B, and user N is connected with node D.Communicate between network M and the network N.2 physical pathways are arranged between network M and the network N, that is: user N ← → node D ← → node C ← → Node B ← → network M, user N ← → node D ← → node E ← → node F ← → network X ← → node A ← → node G ← → Node B ← → user M.

When using the subring resist technology; General Definition ring protection link and Control Node; That is: under the trouble-free situation of ethernet ring network; In the subring data message being blocked the link that prevents data loopback formation is the ring protection link, through the operation to this section ring protection link, can carry out the primary path of subring and the switching in protection path.The node that has the ring protection link is called Control Node (perhaps being called host node) here.Shown in Fig. 2 a, the node that looped network comprises has G, A, B, C, D, E and F, and the link that comprises has < G, A >, < A, B >, < B, C >, < C, D >, < D, E>and < E, F>link.Node A is a Control Node, with its w direct port connection link < A, B>be the ring protection link.

When ring uplink when intact, Control Node is blocked the data message forwarding function of the port that links to each other with the ring protection link, and loop free produces in the network, has prevented " broadcast storm " that cause owing to network loop.Shown in Fig. 2 a, Control Node A has blocked the protected data forwarding capability of e port, and the communication path of user M and N is: user M ← → Node B ← → node C ← → node D ← → user N.

When link occurs fault, Control Node is decontroled the data message forwarding function of port that links to each other with the ring protection link, thereby has ensured professional connection.Shown in Fig. 2 b; < B on the ring; C>fault taken place in link, and Control Node A has decontroled the data message forwarding function of port w, and the communication path that user M and N are new is: user M ← → Node B ← → node A ← → node G ← → network X ← → node F ← → node E ← → node D ← → user N.

When the group ring protection is switched, need to propagate a large amount of control messages, these control messages are in the control channel of subring, to propagate.The control channel of subring has two kinds of configuration modes, and a kind of is the configuration mode that does not have tunnel, that is, the control channel of subring only is configured in the subring (shown in Fig. 3 a).The control channel configuration of another kind of subring comprises part and tunnel (Virtual Channel) in the subring.Tunnel is to be configured in other network or other ring (comprising other subring) between the interconnection point provides the subring of transmission channel for the subring protocol massages control channel (shown in Fig. 3 b).

In ethernet ring network, the node on each looped network is connecting a large amount of subnets, is safeguarding a huge addresses forwarding table.Shown in Fig. 4 a, the link in the subring < A, B>is the ring protection link, and node A is a Control Node.Each Ethernet switching node has connected 10000 sub-net in the subring, and each switching node has safeguarded that a huge single level address transmits, and each transmits the forwarding entry maximum that comprises can reach 5 * 1010.When network topology changes (link switchover that causes like link occurs fault); Each node on the ring is transmitted refresh address; The refresh all that transmit each address of node has not only increased the weight of each address of node learning tasks; And increased the scale of broadcast storm on the ethernet ring network greatly, network performance has been brought great harm.

In fact, when network topology changed, the address of node was transmitted and there is no need on the refresh all ring.Shown in Fig. 5 a, under trouble-free situation, the communication path between B0 and the E0 is: B0 ← → B ← → C ← → D ← → E ← → E0.In other words, Node B, C and D are the MAC Addresss of learning subnet E0 through their w port on the ring, and node E, D and C are the MAC Addresss of learning subnet B0 through their e port.When the link occurs fault on the looped network, link switchover will take place in looped network, shown in Fig. 5 b; Link < C, D>on ring is when breaking down, and node C blocks the data message forwarding of w port; And periodically send SF1 message (fault warning 1 message) along the e port; Node D blocks the data message forwarding of e port, and periodically sends SF2 message (fault warning 2 messages) along the w port, and Control Node A opens the data forwarding function of w port.At this moment, Node B and C should delete the MAC Address of the subnet E0 that they learn on the w port, and node D needn't delete the MAC Address of the subnet E0 that it learns on the w port.Node E and D should delete the MAC Address of the subnet B0 that they learn on the e port, and node C needn't delete the MAC Address of the subnet B0 that it learns on the e port.

Below we do further analysis to this phenomenon.Shown in Fig. 6 a, two end points of fault point and RPL link are divided into two zones with subring and network X, and zone 1 comprises Node B and node C, and zone 2 comprises node G, A, network X, D, E and node F.Under the trouble-free situation of looped network; The node of zone on 1 only can be learnt the MAC Address of the subnet that node or their on the zone 2 link to each other from their w port, and can not learn the MAC Address of the subnet that node or their on the zone 2 link to each other from their e port.Therefore, when link < C, D>when breaking down, as long as the SF1 message that node C sends the MAC Address of the w port association of refresh node, needs only the MAC Address of refresh node e port association in zone 2 in zone 1.Equally, the node of zone on 2 only can be learnt the MAC Address of the subnet that node or their on the zone 1 link to each other from their e port, and can not learn the MAC Address of the subnet that node or their on the zone 2 link to each other from their w port.Therefore, when link < C, D>when breaking down, as long as the SF2 message that node D sends refreshes the MAC of the e port association of the node on the zone 2.

Through last surface analysis; We can further sum up: " if the topology of subring changes; new port blocked can occur in the subring; the port blocked of subring was divided into a plurality of zones with looped network before these ports changed with current topology, and the node on these zones only refreshes the MAC Address of the port association of the MAC Address of before the topology of subring changes, learning other regional node or user.”

Can find out that through analyzing when the topology of subring changed, the address flush scheme of traditional subring can cause large-scale data broadcasting storm.

Summary of the invention

Technical problem: the present invention seeks to provide a kind of regional address refreshing method of subring, reduce the influence of data broadcasting storm, to promoting the performance of subring to the defective that background technology exists.

Technical scheme: the present invention adopts following technical scheme for realizing above-mentioned purpose:

The regional address refreshing method of a kind of subring of the present invention is following:

Subring is divided into two category nodes with node area, that is: a category node and two category nodes;

One category node: receive that the port of protocol massages or current port blocked different nodes are called a category node when port and the current looped network of when last once looped network cause clogging, receiving protocol massages blocks;

Two category nodes: receive that the port of protocol massages or the identical node of current port blocked are called two category nodes when the port of when last once looped network cause clogging, receiving protocol massages blocks with current looped network;

When a category node is received address flush message that a category node sends or when two category nodes were received the address flush message that two category nodes send, node refreshed the addresses forwarding table of the receiving port association that receives said address flush message;

When a category node is received the address flush message that two category nodes send or when two category nodes were received the address flush message of category node transmission, node refreshed the address forwarding entry of another subring port association of the said address flush message of non-reception;

Said address flush message comprises the TP field, 0: representing sending node is a category node; 1: representing sending node is two category nodes;

Node contains a variable PN, receives the label of port of label or present node cause clogging of the port of address flush message when being used for writing down current looped network cause clogging.

Preferably, said node also will utilize ITU-T prior art G.8032 to judge whether looped network change in topology takes place:

When the port of a node in subring received protocol massages, extract < Node_ID, BPR>information; This port compares < Node_ID, the BPR>information in the message and < Node_ID, the BPR>information of originally preserving at this port.If inconsistent, this port is with original preservation < Node_ID, BPR>deletion, and new < Node_ID, BPR>preservation, simultaneously, node determines looped network change in topology has taken place;

Wherein, Node_ID: the node number that sends protocol massages;

BPR:1: east port block (w) 0: western port block (e).

Preferably, said node contains port blocked.

Beneficial effect:

When the topology of subring changed, the blindness address refresh scheme of traditional subring can cause large-scale data broadcasting storm.The present invention is a kind of address flush scheme based on the zone; It can significantly reduce the scope of address flush; Thereby reduce causing harm of data broadcasting storm, assurance ether multi-ring network can restrained in 50ms behind the link switchover, is very significant to the performance that promotes subring.

Description of drawings

Fig. 1 is the structure of subring;

Fig. 2 a is the data forwarding sketch map under the subring non-failure conditions;

Fig. 2 b is the sketch map that the data flow protection is switched under the subring situation about breaking down;

Fig. 3 a and 3b are two kinds of control channels of subring;

Fig. 5 a, 5b and 6a are the sketch mapes that conventional brush new departure goes wrong;

Fig. 7 and 8 is particular flow sheets that the present invention implements;

Fig. 9 a and 9b are embodiments of the invention 1;

Figure 10 a and 10b are embodiments of the invention 2.

Embodiment

The present invention proposes a kind of regional address refreshing method of subring, and subring is divided into two category nodes with node area, that is: a category node and two category nodes.

When one (two) category node was received the address flush message of one (two) category node (containing port blocked) transmission, node refreshed the related addresses forwarding table of receiving port that receives said address flush message;

When one (two) category node was received the address flush message of two (one) category nodes (containing port blocked) transmission, node refreshed the address forwarding entry of another subring port association of the said address flush message of non-reception;

For realizing above-mentioned core content, the present invention should also comprise following content:

The address flush message should comprise the TP field, and 0: representing sending node is a category node.1: representing sending node is two category nodes;

Node contains a variable PN, receives the label of port of label or present node cause clogging of the port of address flush message when being used for writing down current looped network cause clogging;

Node also will utilize ITU-T prior art G.8032 to judge whether looped network change in topology takes place, and specifically describes as follows:

When the port of a node in subring received protocol massages, extract < Node_ID, BPR>information.This port compares < Node_ID, the BPR>information in the message and < Node_ID, the BPR>information of originally preserving at this port.If inconsistent, this port is with original preservation < Node_ID, BPR>deletion, and new < Node_ID, BPR>preservation, simultaneously, node determines looped network change in topology has taken place;

Node_ID: the node number that sends protocol massages;

BPR: only have local meaning.1: east port block (w) 0: western port block (e);

In order to further specify scheme of the present invention, step of the present invention is set forth shown in Fig. 7 a and 8a:

Fig. 7 is the process that node sends band address flush infomational message:

Step 701, looped network generation change in topology produces new choke point;

Step 702, the node that has new port blocked refreshes the address forwarding entry related with this port blocked, checks the variable PN of this node then.

Step 703 judges whether the port numbers of PN record is the same with the port numbers of said port blocked.If the same, change step 704 over to, otherwise change step 705 over to;

Step 704, node is two category nodes with self identification, simultaneously the TP field of the address flush message that sends is put 1, and this message is periodically sent on looped network;

Step 705, node is a category node with self identification, is recorded in the port numbers of said port blocked in the PN variable simultaneously.The TP field of the node address flush message that will send puts 0 in addition, and periodically on looped network, sends this message;

Fig. 8 is the process of the band address flush infomational message received of node processing:

Step 801, node is received protocol massages;

Step 802 judges whether this protocol massages has address flush information, just changes step 803 over to if having address flush information;

Step 803, the receiving port of node read this protocol massages < Node_ID, BPR >, and < Node_ID, BPR>that it and self are preserved compares;

Step 804 judges whether both equate.If unequal, change step 805 over to;

Step 805, this port is preserved < Node_ID, the BPR>of protocol massages, original < Node_ID, the BPR>that self preserves of deletion

Step 806, node read it variable PN, judge whether the port numbers that PN preserves identical with the port numbers that receives this protocol massages.

Step 807 if both are different, changes step 808 over to, otherwise changes step 812 over to;

Step 808, node read the TP field value of said protocol massages.

Step 809 judges whether the TP value is 0.If be 0, change step 810 over to, otherwise change step 811 over to;

Step 810, node refresh the address forwarding entry that receives this protocol massages port association;

Step 811, node refresh the address forwarding entry of another port association of non-this protocol massages of reception;

Step 812, node read the TP field value of said protocol massages.

Step 813 judges whether the TP value is 0.If be 0, change step 814 over to, otherwise change step 815 over to;

Step 814, node refresh the address forwarding entry of another port association of non-this protocol massages of reception;

Step 815, node refresh the address forwarding entry that receives this protocol massages port association;

The situation that embodiment one, subring are broken down:

Shown in Fig. 9 a, the node that looped network comprises has A, B, C, D, E, F and G, and the link that comprises has < G, A >, < A, B >, < B, C >, < C, D >, < D, E>and < E, F>link.Node A is a Control Node, with its w direct port connection link < A, B>be the ring protection link.The w port of node A is the blocking data message forwarding under normal circumstances, and has periodically outwards sent the node cycle protection link obstructions on NR (RB) the message notifying ring along two ring port of subring.The variable PN of node A, B, C, D, E and F writes down port numbers e on the ring.

Shown in Fig. 9 b, fault has taken place in the link of looped network < C, D >, and node C refreshes the MAC Address of w port association, and periodically sends SF1 message (having address flush information) along the e port.Because the port numbers e that the port blocked w of node C and its variable PN preserve is inequality, so the TP field value is set to 0 in the SF1 message, simultaneously, the value of variable PN is updated to w; Node D refreshes the MAC Address of e port association, and periodically sends SF2 message (having address flush information) along the w port.Because the port blocked e of node D is identical with the port numbers e that its variable PN preserves, so the TP field value is set to 1 in the SF2 message;

Node B reads < Node_ID, BPR>in the message after receiving the SF1 message from its w port, finds that < Node_ID, the BPR>that preserve with the w port is inconsistent, and < Node_ID, the BPR>that node is preserved the SF1 message deletes < Node_ID, the BPR>of original preservation.The port numbers e that its variable PN of Node B discovery preserves and it receive that the port numbers of SF1 message is inconsistent, infer it oneself is a category node.Node B reads the TP field of SF1 message, and the value of finding it is 0.Therefore, Node B refreshes the address forwarding entry with the w port association, simultaneously, the value of variable PN is updated to w.

Node A (G) reads < Node_ID, BPR>in the message after receiving the SF1 message from its w port, finds that < Node_ID, the BPR>that preserve with the w port is inconsistent, and < Node_ID, the BPR>that node is preserved the SF1 message deletes < Node_ID, the BPR>of original preservation.Node A (G) finds that the port numbers e that its variable PN preserves receives that with it the port numbers of SF1 message is identical, infers it oneself is two category nodes.Node A (G) reads the TP field of SF1 message, and the value of finding it is 0.Therefore, node A (G) refreshes the address forwarding entry with the e port association.

Node E (F) reads < Node_ID, BPR>in the message after receiving the SF2 message from its e port, finds that < Node_ID, the BPR>that preserve with the e port is inconsistent, and < Node_ID, the BPR>that node is preserved the SF2 message deletes < Node_ID, the BPR>of original preservation.Node E (F) finds that the port numbers e that its variable PN preserves receives that with it the port numbers e of SF2 message is identical, infers it oneself is two category nodes.Node E (F) reads the TP field of SF2 message, and the value of finding it is 1.Therefore, node A (G) refreshes the address forwarding entry with the e port association.

The situation that embodiment two, subring fault disappear

Shown in Figure 10 a; Link < C when looped network; D>fault disappear, node C outwards periodically sends NR (NODE C) (no request message) message along its two intact ports, node D outwards periodically sends NR (NODE D) message along its two intact ports; Node D receives that node C sends NR (NODE C) message; Find that the NODE C in the message is bigger than the node number of oneself, node D opens the forwarding capability of data message of the e port of self, and stops to send NR (NODE D) message.Node A receives the no request message that node C or node D send, and starts WTR timer (Wait Timer).What need particularly point out is: because the NR message does not carry address flush information, so after the last node of ring is received the NR message, can not handle the PN variable of message < Node_ID, BPR>and node.

Shown in Figure 10 b; Behind the WTR timer expiry; Node A blocks the w port; Refresh the MAC Address of its w port association, periodically outwards send NR (RB) 1 message and NR (RB) 2 messages (message has address flush information) respectively, other node " the ring protection link is the blocking data message " in the notice subring along e port and w port.Node A finds that the w of port blocked w and the preservation of PN variable is identical, infers it oneself is two category nodes.Therefore, it puts 1 with the TP field of NR (RB) 1 message and NR (RB) 2 messages.

Node B (C) reads < Node_ID, BPR>in the message after receiving NR (RB) 2 messages from its e port; Find that < Node_ID, the BPR>that preserve with the e port is inconsistent, node is preserved the < Node_ID of NR (RB) 2 messages; BPR >, delete < Node_ID, the BPR>of original preservation.Node B (C) finds that port numbers w that its variable PN preserves and it receives that the port numbers e of NR (RB) 2 messages is inconsistent, infers it oneself is a category node.Node B (C) reads the TP field of NR (RB) 2 messages, and the value of finding it is 1.Therefore, Node B (C) refreshes the address forwarding entry with the w port association, simultaneously, the value of variable PN is updated to e.

Node D (E, F) reads < Node_ID, BPR>in the message after receiving NR (RB) 2 messages from its e port; Find that < Node_ID, the BPR>that preserve with the e port is inconsistent, node is preserved the < Node_ID of SF2 message; BPR >, delete < Node_ID, the BPR>of original preservation.Node D (E, F) finds that the port numbers e that its variable PN preserves receives that with it the port numbers e of NR (RB) 2 messages is identical, infers it oneself is two category nodes.Node E (F) reads the TP field of SF2 message, and the value of finding it is 1.Therefore, node D (E, F) refreshes the address forwarding entry with the e port association.

Node G reads < Node_ID, BPR>in the message after receiving NR (RB) 1 message from its w port; Find that < Node_ID, the BPR>that preserve with the w port is inconsistent, node is preserved the < Node_ID of NR (RB) 1 message; BPR >, delete < Node_ID, the BPR>of original preservation.Node G finds that the port numbers w that its variable PN preserves receives that with it the port numbers w of NR (RB) 1 message is identical, infers it oneself is two category nodes.Node G reads the TP field of NR (RB) 1 message, and the value of finding it is 1.Therefore, node G refreshes the address forwarding entry with the w port association.

Can know that from last surface analysis the present invention can make the subring node avoid unnecessary address deletion well, thereby prevents subring because the broadcast storm that whole address flush of traditional scheme cause has improved network performance greatly.

Certainly; The present invention also can have other various embodiments; Under the situation that does not deviate from spirit of the present invention and essence thereof; Those of ordinary skill in the art can make various corresponding changes and distortion according to the present invention, but these corresponding changes and distortion all should belong to the protection range of the appended claim of the present invention.

Claims

1. the regional address refreshing method of a subring is characterized in that said method is following:

When a category node is received address flush message that a category node sends or when two category nodes were received the address flush message that two category nodes send, node refreshed the address forwarding entry of the receiving port association that receives said address flush message;

Said address flush message comprises the TP field, and when the TP field value was 0, sending node was a category node; When the TP field value was 1, sending node was two category nodes;