CN104378296B - A kind of message forwarding method and equipment - Google Patents

Abstract

The invention discloses a kind of message forwarding method and equipment, in the control plane of big double layer network, each node device by one up to all node devices in big double layer network broadcast tree, the VLAN belonging to local mac address and the MAC Address that new will learn is advertised to other node devices；Node device receives the local mac address of other node devices of other node devices notice from the broadcast tree, and according to the big double layer network forwarding-table item for being in advance based on the big double layer network protocol generation, generates original MAC address forwarding-table item；In the datum plane of big double layer network, after node device receives unicast data message, according to the target MAC (Media Access Control) address of the unicast data message and affiliated VLAN, matched in the original MAC address forwarding-table item, and the unicast data message is directly forwarded and without big double layer network protocol encapsulation by matching the outgoing interface of list item.In the present invention, the L2 unicast traffics forwarding based on big double layer network is realized.

Description

A kind of message forwarding method and equipment

Technical field

The present invention relates to the multicasting technology of the communications field, more particularly to a kind of message forwarding method and equipment.

Background technology

STP is used in three traditional layer data center frameworks（Spanning Tree Protocol, Spanning-Tree Protocol）Come Optimize client to the path of server and support connection redundancy.The reasons such as the constringency performance due to STP, generally STP Not over 100 interchangers of network size.Simultaneously because STP needs to block redundance unit and link, network is also reduced Resource bandwidth utilization rate.It is traditional that data center's scale, bandwidth can not have been met based on STP alternate devices and link scheme Demand, and several seconds convergence time after malfunction occurred to a few minutes of STP protocol, the reliability requirement of data center can not be met.Cause This, it is desirable to be able to there is new technology, while double layer network scale is met, can also make full use of redundance unit and link, Link utilization is lifted, and the convergence time after malfunction occurred of data center can be reduced to submicrosecond even Millisecond.

Virtualization can effectively improve the utilization rate of server, reduce energy resource consumption, reduce the O＆M cost of client, institute Great development has been obtained with virtualization technology.The raising for being not only server utilization brought to data center is virtualized, The also change of the network architecture.Virtual machine（Virtual Machine, VM）Migrating technology can make the computing resource of data center Flexibly allocated, further improve the utilization rate of resources of virtual machine.But before and after virtual machine (vm) migration requirement virtual machine (vm) migration IP and MAC（Media Access Control, medium access control）Address is constant, and this is just needed before and after virtual machine (vm) migration Network is in inside same two layers of domain.Because the scope of customer requirement virtual machine (vm) migration is increasing, even across difference Migration between region, different computer rooms, so that the scope of data center's double layer network is increasing, or even there are big rule Mould double layer network（Hereinafter referred to as big double layer network）This frontier special topic.

Tunneling technique can realize big two layers of net in the data center for covering multiple access switch and core switch Network.The Typical Representative of tunneling technique is TRILL（TRansparent Interconnection of Lots of Links are more Link transparency interconnects）And SPB（Shortest Path Bridging, shortest path bridge）Technology, both of which are by borrowing IS- IS（Intermediate System to Intermediate System, Intermediate System-to-Intermediate System）The meter of Routing Protocol Calculation and forward mode, realize the extensive extension of double layer network.These technologies are currently in standardisation process, while tradition is handed over Changing planes not only needs software upgrading, it is also necessary to hardware supported.

In existing big two layers of networking technology, L2 unicasts/broadcast traffic needs to forward after carrying out big two-layer protocol encapsulation, when big Two layers of networking technology and other technologies coexist it is higher to chip requirement during networking, such as 802.1Br technologies and Trill technology co-existence groups During net, message is encapsulated when entering Trill networks from 802.1Br, it is necessary to first remove 802.1Br E-tag encapsulation, then change into Trill is encapsulated；, it is necessary to remove Trill encapsulation when reversely entering 802.1Br networks from Trill networks, then change E-tag encapsulation into. Currently commercial chip is not supported to handle above flow in a pipeline, i.e., can not be first to message in a pipeline Remove E-tag encapsulation（Trill is encapsulated）, then change Trill encapsulation into（E-tag is encapsulated）Flow, limit Trill/802.1Br Etc. the actual deployment of technology.

In the prior art, when networking coexists in big two layers of networking technology and other technologies, She Ge Shang chip are typically employed in The mode of winding mouth, a kind of complete flow of encapsulation of addition/deletion is fed again into chip through winding mouth and carries out after-treatment, is added Add/delete another encapsulation.Wherein, 802.1Br technologies can be such as Fig. 1 institutes with handling process during Trill technology co-existence networkings Show.

In the prior art, processing mode when big two layers of networking technology and other technologies coexist networking is, it is necessary to E Wai Pei Ge Winding mouth, device port is wasted, while winding mouth turns into the forwarding bottleneck of system, performance is low, and delay is big.

The content of the invention

The embodiment of the present invention provides a kind of message forwarding method and equipment, to make big two-layer network device forwarding L2 mono- Broadcast/broadcast traffic.

In order to reach object above, the embodiments of the invention provide a kind of message forwarding method, is set applied to multiple nodes In the standby big double layer network set up based on big double layer network agreement, this method includes：

Each node device by one up to all node devices in the big double layer network broadcast tree, will newly learn To local mac address and affiliated VLAN be advertised to other node devices, the local mac address includes connecting from this equipment The MAC Address that learns of user network side；

The node device from it is described broadcast tree on receive other node devices notice described in other node devices newly learn After the first MAC Address and affiliated first VLAN practised, identified according to the big double layer network of other node devices, inquiry The big double layer network forward table of the big double layer network protocol generation is in advance based on, it is determined that going out corresponding to other described node devices Interface, and then original MAC address forwarding-table item corresponding to first MAC Address is generated, wherein, the original MAC address turns VLAN in forwarding list item is the first VLAN, target MAC (Media Access Control) address is first MAC Address, outgoing interface is other described sections Outgoing interface corresponding to point device；

After the node device receives unicast data message, according to the target MAC (Media Access Control) address of the unicast data message and Affiliated VLAN, matched in the original MAC address forwarding-table item, and the outgoing interface by matching list item directly forwards institute State unicast data message and without big double layer network protocol encapsulation.

Wherein, the big double layer network mark of other node devices described in the basis, inquiry is in advance based on described big two layers The big double layer network forward table of procotol generation, it is determined that outgoing interface corresponding to other described node devices, including：

Destination node equipment is identified as with the big double layer network of other node devices, forwarded from the big double layer network In table, forwarding-table item corresponding to the big double layer network mark of other node devices is found；

If only including an outgoing interface in the forwarding-table item found, the outgoing interface that will be found, as it is described other Outgoing interface corresponding to node device；

If the forwarding-table item found includes more than two outgoing interfaces, more than two outgoing interfaces are formed one Individual aggregation port, and using the aggregation port as outgoing interface corresponding to other described node devices.

Wherein, the outgoing interface by matching list item directly forwards the unicast data message and without big two layers of net Network protocol encapsulation, including：

If the outgoing interface for matching list item is aggregation port, pass through a member of aggregation port described in HASH algorithms selections Port, and by selected member port, directly forward the unicast data message and sealed without big double layer network agreement Dress.

Wherein, in addition to：

Each node device is also set by the broadcast, and the local mac address that aging occurs and affiliated VLAN are noticed Give other node devices；

Node device receives from the broadcast tree to be occurred always described in the notice of other node devices at other node devices The second MAC Address and affiliated VLAN changed, and from the original MAC address forwarding-table item, delete second MAC Address And the forwarding-table item corresponding to affiliated VLAN.

Wherein, each node device is directed to the combination of itself different interface and VLAN, Fen Bie Pei Ge mono- corresponding MAC Address pool, for being maintained in the local mac address for belonging to corresponding VLAN learnt on the corresponding interface；

Each node device is when local mac address is arrived in new study, if not including the new study in corresponding MAC Address pond During the local mac address arrived, then the new local mac address learnt is increased in MAC Address pond, and, in local mac After one section of preset time of aging occurs for location, the local mac address of aging is deleted from corresponding MAC Address pond；

Each node device, when the local mac address newly learnt is further added in MAC Address pond, pass through institute Broadcast tree is stated, the new local mac address learnt and affiliated VLAN are advertised to other node devices；And in MAC Address When the local mac address of aging is deleted in pond, set by the broadcast, by the local mac address of aging and affiliated VLAN It is advertised to other node devices.

Wherein, this method also includes：

The She Ge VLAN broadcast domain scope Shang node device, wherein, the broadcast domain scope is included on the node device Allow the big double layer network side joint cut positioned at the VLAN on tree in the user-side port and the node device that the VLAN passes through Mouthful, the cutting tree is obtained based on the VLAN beta prunings for being distributed tree to the broadcast tree；

After node device receives data message to be broadcast, the affiliated VLAN of data message to be broadcast is determined, and at this On the port of other in addition to receiving port included by VLAN broadcast domain scope, directly forward the data message to be broadcast and Without big double layer network protocol encapsulation, wherein, the data message to be broadcast includes unknown unicast message, broadcasting data messages And unknown multicast packets.

Wherein, in the network topology change of the big double layer network, big double layer network forward table and/or VLAN is caused to cut After tree updates, this method also includes：

When the outgoing interface of forwarding-table item of the node device in big double layer network forward table updates, renewal is corresponding Original MAC address forwarding-table item outgoing interface；

When forwarding-table item of the node device in big double layer network forward table is deleted, it is determined that deleted forward table The destination node equipment of item and the local mac address for belonging to the destination node equipment, and then delete the destination node equipment Local mac address corresponding to original MAC address forwarding-table item；

The node device cuts tree, Chong Xin She Ge couple after VLAN cutting trees update according to the VLAN after renewal Answer VLAN broadcast domain scope.

The embodiment of the present invention additionally provides a kind of big two-layer network device, is applied to multiple node devices as node device The big double layer network set up based on big double layer network agreement, the big two-layer network device are included：

Address management module, for by one up to all node devices in the big double layer network broadcast tree, will Newly the local mac address that learns and affiliated VLAN are advertised to other node devices to the big two-layer network device, described Ground MAC Address includes the MAC Address learnt from the user network side of this equipment connection；

List item maintenance module, lead to for receiving other node devices from the broadcast tree in the big two-layer network device After the first MAC Address and affiliated first VLAN that other the described node devices accused newly learn, according to other described nodes The big double layer network mark of equipment, inquiry are in advance based on the big double layer network forward table of the big double layer network protocol generation, really Outgoing interface corresponding to other fixed described node devices, and then generate original MAC address forward table corresponding to first MAC Address , wherein, the VLAN in the original MAC address forwarding-table item is the first VLAN, target MAC (Media Access Control) address is described first MAC Address, outgoing interface are outgoing interface corresponding to other described node devices；

Packet forwarding module, for after the big two-layer network device receives unicast data message, according to the list The target MAC (Media Access Control) address of multicast data message and affiliated VLAN, matched, and passed through in the original MAC address forwarding-table item Matching list item outgoing interface directly forward the unicast data message and without big double layer network protocol encapsulation.

Wherein, the list item maintenance module is specifically used for, and is accomplished by the following way according to other node devices Big double layer network mark, inquiry is in advance based on the big double layer network forward table of the big double layer network protocol generation, it is determined that described Outgoing interface corresponding to other node devices：

Destination node equipment is identified as with the big double layer network of other node devices, forwarded from the big double layer network In table, forwarding-table item corresponding to the big double layer network mark of other node devices is found；

If only including an outgoing interface in the forwarding-table item found, the outgoing interface that will be found, as it is described other Outgoing interface corresponding to node device；

If the forwarding-table item found includes more than two outgoing interfaces, more than two outgoing interfaces are formed one Individual aggregation port, and using the aggregation port as outgoing interface corresponding to other described node devices.

Wherein, the packet forwarding module is specifically used for, and is accomplished by the following way straight by the outgoing interface for matching list item Switch through and send out unicast data message described and without big double layer network protocol encapsulation：

If the outgoing interface for matching list item is aggregation port, pass through a member of aggregation port described in HASH algorithms selections Port, and by selected member port, directly forward the unicast data message and sealed without big double layer network agreement Dress.

Wherein, the address management module is additionally operable to, by it is described broadcast set, by occur aging local mac address with And affiliated VLAN is advertised to other node devices；

The list item maintenance module is additionally operable to, and other sections described in the notice of other node devices are received from the broadcast tree The second MAC Address of aging and affiliated VLAN occur at point device, and from the original MAC address forwarding-table item, deletes Second MAC Address and the forwarding-table item corresponding to affiliated VLAN.

Wherein, the big two-layer network device be directed to itself different interface and VLAN combination, Fen Bie Pei mono- correspondence of Ge MAC Address pond, for being maintained in the local mac address for belonging to corresponding VLAN learnt on the corresponding interface；

The address management module is specifically used for, when local mac address is arrived in new study, if in corresponding MAC Address pond not Include this it is new learn local mac address when, then the new local mac address learnt is increased in MAC Address pond, with And after one section of preset time that aging occurs for local mac address, by the local mac address of aging from corresponding MAC Address pond Delete；When the local mac address newly learnt is added in MAC Address pond, set by the broadcast, the sheet that new will learn Ground MAC Address and affiliated VLAN are advertised to other node devices；And the local mac of aging is deleted in MAC Address pond During address, set by the broadcast, the local mac address of aging and affiliated VLAN are advertised to other node devices.

Wherein, the packet forwarding module is additionally operable to,

The She Ge VLAN broadcast domain scope Shang node device, wherein, the broadcast domain scope is included on the node device Allow the big double layer network side joint cut positioned at the VLAN on tree in the user-side port and the node device that the VLAN passes through Mouthful, the cutting tree is obtained based on the VLAN beta prunings for being distributed tree to the broadcast tree；

After receiving data message to be broadcast, the affiliated VLAN of data message to be broadcast is determined, and in the wide of the VLAN Broadcast on the port of other in addition to receiving port included by the scope of domain, directly forward the data message to be broadcast and without big Double layer network protocol encapsulation, wherein, the data message to be broadcast includes unknown unicast message, broadcasting data messages and unknown group Report text.

Wherein, the list item maintenance module is additionally operable to, and in the network topology change of the big double layer network, is caused big two layers After forwarded table updates, if the outgoing interface of the forwarding-table item in big double layer network forward table updates, renewal is corresponding Original MAC address forwarding-table item outgoing interface；If the forwarding-table item in big double layer network forward table is deleted, it is determined that being deleted Forwarding-table item destination node equipment and belong to the local mac address of the destination node equipment, and then delete the purpose Original MAC address forwarding-table item corresponding to the local mac address of node device；

The packet forwarding module is additionally operable to, and in the network topology change of the big double layer network, causes VLAN to cut tree After updating, tree is cut according to the VLAN after renewal, Chong Xin She Ge correspond to VLAN broadcast domain scope.

In the above embodiment of the present invention, each node device is set by one up to all nodes in the big double layer network Standby broadcast tree, the new local mac address learnt and affiliated VLAN are advertised to other node devices；The node device Received from the broadcast tree new the first MAC Address learnt of other node devices described in the notice of other node devices and After affiliated first VLAN, identified according to the big double layer network of other node devices, inquiry is in advance based on big two layers of net The big double layer network forward table of network protocol generation, it is determined that outgoing interface corresponding to other described node devices, and then generate described the Original MAC address forwarding-table item corresponding to one MAC Address；After the node device receives unicast data message, according to described The target MAC (Media Access Control) address of unicast data message and affiliated VLAN, matched, and led in the original MAC address forwarding-table item The outgoing interface of overmatching list item directly forwards the unicast data message without big double layer network protocol encapsulation, realizes base Forwarded in the L2 unicast traffics of big double layer network, moreover, when L2 unicast traffics forward in big double layer network, it is big by two without carrying out Layer network protocol encapsulation.

Brief description of the drawings

Fig. 1 is 802.1Br technologies in the prior art and handling process schematic diagram during Trill technology co-existence networkings；

Fig. 2 is a kind of configuration diagram of Trill networkings provided in an embodiment of the present invention；

Fig. 3 is the schematic diagram of VLAN broadcast domains in a kind of Trill networkings provided in an embodiment of the present invention；

Fig. 4 is a kind of schematic flow sheet of two layers of unicast message retransmission method provided in an embodiment of the present invention；

Fig. 5 is a kind of schematic diagram of Trill networkings link failure provided in an embodiment of the present invention；

Fig. 6 is a kind of schematic diagram of Trill networkings interior joint equipment fault provided in an embodiment of the present invention；

Fig. 7 is a kind of structural representation of big two-layer network device provided in an embodiment of the present invention.

Embodiment

In message retransmission technique provided in an embodiment of the present invention, in the control plane of big double layer network, each node device By one up to all node devices in big double layer network broadcast tree, by the new local mac address learnt and the MAC VLAN belonging to location is advertised to other node devices；Node device receives other sections of other node devices notice from the broadcast tree The local mac address of point device, and according to the big double layer network forwarding-table item for being in advance based on the big double layer network protocol generation, it is raw Into original MAC address forwarding-table item；In the datum plane of big double layer network, after node device receives unicast data message, root Target MAC (Media Access Control) address and affiliated VLAN according to the unicast data message, are matched in the original MAC address forwarding-table item, and Outgoing interface by matching list item directly forwards the unicast data message and without big double layer network protocol encapsulation.

Wherein, the local mac address of node device includes the node device from the user network side that the node device connects The MAC Address learnt；The target MAC (Media Access Control) address of the original MAC address forwarding-table item of node device is other above-mentioned node devices Local mac address, outgoing interface is that the outgoing interfaces of other node devices is gone in above-mentioned big double layer network forwarding-table item.

Specifically, in embodiments of the present invention, when any node equipment receives other node devices in big double layer network New the first MAC Address learnt of other node devices of notice and during an affiliated VLAN, according to other nodes The big double layer network mark of equipment, inquiry are in advance based on the big double layer network forward table of big double layer network protocol generation, it is determined that should Outgoing interface corresponding to other node devices, and then generate original MAC address forwarding-table item corresponding to first MAC Address.

Wherein, node device determines that outgoing interface can specifically be accomplished by the following way corresponding to other node devices：

Node device from big double layer network forward table, can find the big double layer network mark of other node devices Corresponding forwarding-table item；If only including an outgoing interface in the forwarding-table item searched, the outgoing interface that will be found, other are used as Outgoing interface corresponding to node device；If the forwarding-table item found includes more than two outgoing interfaces, by the two or more Outgoing interface form an aggregation port, and using the aggregation port as outgoing interface corresponding to other node devices.Wherein, it is above-mentioned More than two outgoing interfaces that forwarding-table item includes can be with equal cost multipath（ECMP）Form exist.

Correspondingly, when node device needs to forward unicast data message by the outgoing interface for matching list item, if matching list The outgoing interface of item is aggregation port, then can pass through HASH（Hash）One member port of the algorithms selection aggregation port, and By selected member port, the unicast data message is directly forwarded and without big double layer network protocol encapsulation.

Further, in embodiments of the present invention, when the local mac address aging of node device, node device also needs The local mac address that aging occurs and affiliated VLAN are advertised to other node devices, is deleted and sent out by other node devices Forwarding-table item corresponding to the MAC Address and affiliated VLAN of raw aging.

Specifically, each node device can be set by broadcasting, by the local mac address that aging occurs and affiliated VLAN It is advertised to other node devices；When node device receives the MAC Address of generation aging at other node devices from broadcast tree （Second MAC Address）And during affiliated VLAN, from original MAC address forward table, delete second MAC Address and affiliated Forwarding-table item corresponding to VLAN.

In the embodiment of the present invention, node device can be carried out among the nodes in MAC address aging by notification packet MAC Address is synchronous.The situation for MAC Address vacation aging occur is possible in practical application.So-called MAC Address vacation aging, refers to end After end data sends message, one section of long period is spaced（More than the MAC address aging time）Without retransmiting message, then again Message is sent, repeatedly in this way, causing the MAC Address of the terminal to be learnt on the port of node device, aging, learning again, again Aging ..., each of which time aging or learning process can all trigger the transmission of notification packet.

In order to reduce the MAC Address notification packet quantity forwarded that frequently false aging is triggered, mitigate at the CPU of node device Reason burden, each node device can be directed to the combination of itself different interface and VLAN in the present embodiment, Fen Bie a pair of Pei Ge The MAC Address pond answered, for being maintained in the local mac address for belonging to corresponding VLAN learnt on the corresponding interface；

Each node device arrives local mac address in new study, such as records what is newly learnt in MAC forwarding-table items After ageing timer corresponding to MAC Address and foundation, if not including the new local mac learnt in corresponding MAC Address pond During address, then the new local mac address learnt is increased in MAC Address pond.Occur one section of aging in local mac address After preset time, just the local mac address of aging is deleted from corresponding MAC Address pond, that is to say, that in MAC forwarding-table items In MAC Address because of ageing timer time-out and after aging, after one section of preset time, just by the MAC Address from corresponding MAC Deleted in pond location.

Accordingly, each node device, when the local mac address newly learnt is further added in MAC Address pond, Set by the broadcast, the new local mac address learnt and affiliated VLAN are advertised to other node devices；And When the local mac address of aging is deleted in MAC Address pond, set by the broadcast, by the local mac address of aging and institute Category VLAN is advertised to other node devices.That is, the MAC address aging in MAC forwarding-table items can't directly trigger it is logical The transmission of message is accused, and is to wait for after the MAC Address deletes from MAC Address pond, just goes to send corresponding notification packet.

Below by taking the Trill technologies in data center as an example, the message retransmission technique side that is provided the embodiment of the present invention Case is described in detail, and relative theory is equally applicable to other big two layers of networking technologys, such as Vxlan（virtual Extended VLAN, virtual extended VLAN（Virtual Local Area Network, VLAN）, IETF draft （Draft））, SPB technologies etc..

Using the big double layer network of the data center of Trill technique constructions, core layer, Access Layer are generally divided into.Generally, run The equipment of Trill agreements is referred to as RBridge（That is Routing Bridge, there is the bridge equipment of routing forwarding characteristic, also referred to as Routing bridge, hereinafter referred to as RB）, RB composition Trill domains（That is Trill networks）.RB recognizes Trill by running link-state protocol The topology in domain, and the nickname of other each RB using the generation of shortest path tree algorithm out of the RB arrival Trill networks Route forwarding table（Referred to as Trill routing tables, including nickname unicast routing tables and multicast table）.

In embodiments of the present invention, each RB carries out nickname router-level topologies using Trill control protocols in Trill domains When, the Trill routing tables of generation only retain in control plane, without issuing data forwarding plane（Retain on software, without Issue hardware）.

Accordingly, due to there is no nickname forwarding-table items in the hardware forwarding table of each RB in Trill domains, therefore, when Each RB uses ESADI in Trill domains（End Station Address Information, for terminal MAC address each The agreement of active synchronization between RB）When protocol massages to each RB local mac address mutually notice and be synchronous, ESADI associations View message needs to be sent to CPU progress software forwards and processing.Wherein it is possible to a Trill is previously generated on the control plane Broadcast tree, do not do any cutting, all nodes（Each RB in Trill domains）It is reachable, it is specifically used to forward ESADI protocol massages.

It should be noted that for the automatic study of MAC Address（Hardware learns source MAC）, table of its study of one side The not necessarily correct list item of item, easily forms loop and causes broadcast storm, and on the other hand, MAC Address learnt automatically In journey message not necessarily can transmitted to CPU, lead to not by corresponding MAC Address by ESADI protocol massages be synchronized to distal end set It is standby.

For example, the source MAC of a message is 0-0-1, target MAC (Media Access Control) address 0-0-2, if being not prohibited by MAC Address certainly Dynamic study, RB search MAC table after port1 receives the message, and discovery has list item corresponding to 0-0-2, points to port2.If It is not prohibited by MAC Address automatically to learn, RB hardware will learn the corresponding relation of source MAC and receiving port automatically, directly add Add 0-0-1->Port1 MAC forwarding-table items, are then directly forwarded message by port2, in this process, will not be by message The CPU of transmitted to CPU, i.e. RB does not perceive to this, leads to not synchronize the MAC Address by ESADI protocol massages.

Based on this, in embodiments of the present invention, for all user-side ports in Trill domains, it is necessary to forbid MAC to learn automatically Practise（Hardware is forbidden to learn）, being changed to software learning, either Shou Gong Pei Ge MAC Address or agreement automatically generate original MAC address turn The mode of forwarding list item.By taking software learning as an example, after message entry port, CPU is first gone up, the source MAC of message is analyzed by CPU, MAC address entries are added to hardware.Wherein, on message after CPU, CPU can extract the source MAC of message, turn except addition is local Outside forwarding list item, it can also be synchronized by address synchronization mechanism to other distant-end nodes.

And the port for allowing Trill messages to pass through, such as Trill mouths and Hybrid（Mix）Mouthful, implement in the present invention In example, then need to forbid its study（Learn including software, hardware）MAC Address, is changed to Shou Gong Pei Ge or agreement automatically generates The mode of original MAC address forwarding-table item.For example, it is necessary to according to topology after receiving the MAC Address that distant-end node diffuses through Information calculates this end node and arrives the path of the MAC Address, if existing a plurality of, needs to add an ECMP list item, such as the MAC Address passes through link（Link）1 is diffused into this end node, but passes through and calculate, and finds link1, and 2,3 be all to reach the MAC Address Path, then needed when descending forwarding list item link1, the exit port after 2,3 polymerizations as the arrival MAC Address.

In addition, during using ESADI protocol synchronization MAC Address, all nodes（Including transmission node（That is transit nodes）） It is required for handling ESADI messages.

In Trill domains each RB receive other RB transmission ESADI protocol massages after, extract wherein carry MAC Address, The nickname information of vlan information and other RB, and the nickname routing tables according to corresponding to nickname inquiries, Exit port corresponding to the nickname is determined, then outgoing interface, the MAC Address and vlan information that are extracted determined by, Generation original MAC address forwarding-table item simultaneously issues data forwarding plane.

Specifically, can corresponding to the MAC+VLAN that is carried in ESADI protocol massages is replaced nickname route table items In nickname, if the exit port of the nickname route table items is single exit port, exit port in corresponding MAC address entries Directly use the exit port of the nickname route table items；If the nickname route table items have equal-cost route, corresponding MAC Exit port in list item can be each port in the nickname route table items（Port corresponding to each equal-cost route）It is corresponding poly- Close port.

When the data message that each RB forwardings receive in Trill domains, not to message progress Trill encapsulation, and according to Corresponding original MAC forwarding-table items directly forward to message；If the MAC inquired according to the target MAC (Media Access Control) address of message turns When the exit port of forwarding list item is single port, then the message is forwarded from the exit port；If according to the target MAC (Media Access Control) address of message The exit port of the MAC forwarding-table items inquired is aggregation port, then can be shared by the hash of corresponding aggregated links, selection Port corresponding to certain sub-links of aggregated links forwards to the message.

It should be noted that in embodiments of the present invention, the difference of aggregation port and conventional polymeric port in MAC address entries Part is：The equipment at traditional aggregated links both ends is required for Pei Ge to polymerize, and the aggregated links in the embodiment of the present invention is only Need to be aggregated links from the point of view of a side apparatus, opposite end not necessarily needs to be aggregated links.

It can be seen that when the node device of big double layer network receives two layers of unicast message, be not required to by above flow Big two-layer protocol encapsulation is carried out to two layers of unicast message, but according to the target MAC (Media Access Control) address of two layers of unicast message hard On part inquiry corresponding to MAC forwarding-table items, and according to the MAC forwarding-table items inquired by two layers of unicast message with native （It is primary）Form forwards to the message, when networking coexists in big two layers of networking technology and other technologies, with 802.1Br technologies Exemplified by Trill technology co-existence networkings, message from 802.1Br networks enter Trill networks when, it is only necessary to remove E-tag encapsulation, Without carrying out Trill encapsulation, you can realize forwarding of the message in Trill networks；Reversely enter from Trill networks During 802.1Br networks, because message does not carry out Trill encapsulation, therefore, the E-tag that 802.1Br need to be only carried out to the message is sealed Dress, you can realize forwarding of the message in 802.1Br networks, effectively having evaded in chip can not locate simultaneously in a pipeline Reason first removes E-tag encapsulation（Trill is encapsulated）, then change Trill encapsulation into（E-tag is encapsulated）Flow the defects of, be not required to volume On the premise of Wai Pei Ge winding mouths, the message forwarding that networking coexists in big two layers of networking technology and other technologies is realized.

In order to illustrate the embodiments of the present invention more clearly, below by taking the Trill group-network constructions shown in Fig. 2 as an example, this is illustrated The specific implementation process for the message retransmission technique that inventive embodiments provide.

As shown in Fig. 2 be a kind of configuration diagram of Trill networkings provided in an embodiment of the present invention, wherein：

Each access layer equipment（That is access nodes）Nickname be respectively nickname-1, nickname-2 ..., nickname-n；

Each core layer equipment（That is core nodes）Nickname be respectively nickname-g1, nickname-g2, Nickname-g3 and nickname-g4；

MAC Address on each access layer equipment is as follows：

nickname-1（That is nickname is nickname-1 equipment, similarly hereinafter）On have VLAN1/mac1, VLAN2/ mac2、VLAN3/mac3；

There are VLAN1/mac4, VLAN2/mac5, VLAN3/mac6 on nickname-4；

There is VLAN10/mac10 on nickname- (n-1)；

There are VLAN1/mac7, VLAN2/mac8, VLAN3/mac9 on nickname-n；

L3 gateways have VLAN1, VLAN2 and VLAN3 in each core layer equipment in each core layer equipment.With Exemplified by nickname-g1, the MAC Address of L3 interfaces thereon is：VLAN1/mac-v1、VLAN2/mac-v2、VLAN3/mac- v3。

Jun Pei Ge have L2ISIS agreements on all RB in Trill domains, and each RB calculates Trill domains by L2ISIS agreements Nickname unicast routing tables and multicast table, and the nickname unicast routing tables and multicast table only retain on software, not under Issue hardware.Wherein, in networking shown in Fig. 2, the unicast routing table to other nodes of part of nickname-1 generations can be as Shown in table 1：

Table 1

Destination（Purpose）	Nexthop（Next-hop）
		Nickname-4	P1、P2、P3、P4（ECMP）
Nickname-g1	P1
		Nickname-n	P1、P2、P3、P4（ECMP）

Wherein, there is a plurality of equal-cost route, i.e. ECMP in nickname-1 to nickname-4 singlecast router（Equal- Cost Multipath Routing, equal cost multipath）.

Hai Pei Ge have ESADI agreements on all RB in Trill domains, by the synchronization mechanism of ESADI agreements, will be used under each RB The MAC Address of family side is synchronized to every other RB（Including transmission node）.Wherein, RB receive after EASDI protocol massages, it is necessary to By the ESADI protocol massages transmitted to CPU, forwarded by CPU according to the nickname multicast tables preserved in software.

It should be noted that due to transmission node in existing Trill agreements（Transit nodes）In be only stored with Nickname route table items, without MAC address entries, therefore, transmission node need not handle ESADI associations in existing Trill networks Message is discussed, but in embodiments of the present invention, because each RB needs to generate the MAC forwarding-table items for being used for instructing message to forward, because This, transmission node is also by other RB in ESADI protocol massages and Trill domains（Core layer equipment and access layer equipment）It is synchronous Mac address information.

Meanwhile RB receives ESADI protocol massages, and by after the ESADI protocol massages transmitted to CPU, it is necessary to according to this The MAC Address carried in ESADI protocol massages and corresponding nickname information, and the nickname unicasts preserved in software Routing table, MAC forward tables corresponding to generation, and it is issued to hardware.

Wherein, arrived when the nickname information inquiries according to corresponding to the MAC Address carried in EASID protocol massages When exit port is single physical port in nickname singlecast router list items, the MAC forward tables using the physical port as generation In to should MAC Address exit port, and issue hardware.

For example, nickname-1 receives the VLAN1/mac- that nickname-g1 is synchronously come by ESADI protocol massages V1, after the ESADI protocol massages transmitted to CPU, nickname unicast routing tables are inquired about, are inquired corresponding to nickname-g1 The exit port of nickname singlecast router list items is P1（Referring to table 1）, MAC forwarding-table items as shown in table 2 may finally be generated：

Table 2

VLAN	MAC	Egress port（Exit port）
			VLAN1	mac-v1	P1

As the nickname that the nickname information inquiries according to corresponding to the MAC Address carried in ESADI protocol massages are arrived When exit port is ECMP in singlecast router list item, each subport is polymerize, as in the MAC forward tables of generation to should MAC The exit port of address, and issue hardware.

For example, nickname-1 receives the VLAN1/mac4 that nickname-4 is synchronously come by ESADI protocol massages, will After the ESADI protocol massages transmitted to CPU, nickname singlecast router list items are inquired about, are inquired corresponding to nickname-4 The exit port of nickname singlecast router list items is that P1, P2, P3 and P4 form ECMP（Referring to table 1）.Now in nickname-1 P1-P4 is done in equipment and polymerize, an aggregation port is formed, MAC forwarding-table items as shown in table 3 may finally be generated：

Table 3

VLAN	MAC	Egress port（Exit port）
			VLAN1	mac4	Trunk（P1、P2、P3、P4）

Further, it is contemplated that technical scheme provided in an embodiment of the present invention is needed by software synchronization MAC Address, for Big MAC environment（Such as more than 1M scales）, it is necessary to which synchronous MAC Address amount is larger, the processing load of ESADI protocol massages is heavier, In embodiments of the present invention, can take some measures to reduce the processing load of ESADI protocol massages.

When ESADI agreements just bring into operation, each RB synchronously respective MAC Address when, more ESADI agreement reports can be sent Text, because the process belongs to the initial transition process that network just established, the duration is shorter, therefore, although should during it is each RB processing loads are heavier, but overall performance of network are influenceed smaller.

When the initial MAC Address of network synchronously completes, long-term Timing Synchronization is put into and process that triggering refreshes, this Process is the main existence of network, if synchronous MAC Address amount is excessive, overall performance of network is had a great influence.Therefore, exist In the embodiment of the present invention, it can take some measures to reduce in Timing Synchronization and triggering refresh process, ESADI protocol massages Processing load.

Specifically, for big MAC environment, in Timing Synchronization and triggering refresh process, overall performance of network is influenceed Principal element has two：

1）, Timing Synchronization MAC Address amount it is big, ESADI protocol massages traffic volumes are big, cause in network RB to handle load and add Weight.For the problem, in embodiments of the present invention, the time of timing synchronization interval according to the virtual condition of network, can be extended, For example, the temporal frequency of the Timing Synchronization defined in standard is 30 seconds synchronous 3 times, in embodiments of the present invention, then Ke Yi She Ge For 1 minute（Or the longer time）It is synchronous 3 times, to reduce sync message transmission frequency, reduce RB processing loads.

2）, MAC Address frequently false aging（For example, the next main frames of RB send message into Trill domains, MAC Address is learnt To corresponding RB, afterwards, the main frame long period does not retransmit message, causes corresponding RB to the MAC Address of the main frame learnt Aging is carried out, afterwards, the main frame sends message into Trill domains again, and corresponding RB needs with relearning the MAC of the main frame Location）, cause ESADI renewal message frequently to send.For the problem, in embodiments of the present invention, can at least use following Two ways reduces the influence caused by the problem to network performance：

a）, extend the MAC address aging time（That is the timing length of MAC address aging timer）, reduce concussion number；

b）, in ESADI agreements introduce a new timer Ta（The timing length of the Ta is determined more than MAC address aging When device timing length, MAC ageing timers and Ta are reset simultaneously during MAC address learning, Ta timing length can basis Concrete application scene and MAC Address scale Lai She Ge）, only when Ta time-out, MAC Address does not refresh yet, just to the MAC Address Carry out aging（I.e. delete MAC forward tables in should MAC Address list item, similarly hereinafter）, ESADI revocation messages are externally sent, are removed Sell the MAC Address learnt on other RB（I.e. other RB equipment delete self MAC forward table in should MAC Address table ）.

For mode b）, in embodiments of the present invention, can be using current all MAC Address under the+VLAN of port as one Individual pool（Pond）, only pool changes, just external to send LSP refreshing MAC Address.Wherein, pool change principle can be with Including：

（1）, new MAC Address occurs in local pool for the first time（Newly local mac address is arrived in study to RB）When, it is believed that Pool changes, and sends ESADI renewal messages and learns the new MAC Address by other RB；

（2）During Ta time-out, the MAC Address learnt does not refresh also, then it is assumed that the MAC Address needs aging, deletes this List item corresponding to the MAC Address on ground software（List item corresponding to the MAC Address on hardware surpasses in corresponding ageing timer Constantly delete）, while externally send ESADI renewal messages and aging is carried out to the MAC Address by other nodes；

（3）, from other RB receive local pool（Receive the RB of ESADI renewal messages local pool）Middle MAC The ESADI renewal messages that location is refreshed（VMotion scenes, i.e. virtual machine (vm) migration scene）, refresh local software on the MAC Address The exit port of corresponding list item, also on synchronous refresh hardware to should MAC Address list item exit port.

The indeclinable principles of pool can include：

（1）, MAC address aging timing time-out, but when Ta has not timed out, the MAC Address does not refresh yet, then only deletes local List item corresponding to the MAC Address on hardware, but corresponding list item still retains on software, does not send ESADI renewal messages externally；

（2）, the MAC Address that is removed on hardware is when learning to hardware again, due to now also retaining on software List item corresponding to the MAC Address, therefore, it is not considered as pool and changes, does not send ESADI renewal messages externally.

By the above-mentioned means, MAC Address can be avoided, frequently ESADI caused by false aging updates that message traffic volume is big to ask Topic, RB processing loads are reduced, improve the stability of network.

For broadcasting packet, in embodiments of the present invention, can use according to chip VLAN broadcast domain scopes（bitmap） The mode directly broadcasted is forwarded.

In embodiments of the present invention, can Shang node device She Ge VLAN broadcast domain scope, wherein, the broadcast domain model Enclose including allowing the user-side port that the VLAN passes through to cut tree positioned at the VLAN with the node device on the node device On big double layer network side interface, the VLAN, which cuts tree, to be obtained based on the VLAN beta prunings for being distributed tree to the broadcast tree；

After node device receives data message to be broadcast, the affiliated VLAN of data message to be broadcast is determined, and at this On the port of other in addition to receiving port included by VLAN broadcast domain scope, directly forward the data message to be broadcast without Big double layer network protocol encapsulation is carried out, wherein, the data message to be broadcast includes unknown unicast message, broadcasting data messages and not Know multicast message

Specifically, still by taking the Trill technologies in data center as an example, in embodiments of the present invention, can be according to TRILL domains VLAN cut the broadcast domain scope of each VLAN in tree Lai She Ge hardware chips, user-side port on port according to allowing to pass through VLAN carry out Dui Ying She Ge；For TRILL mouths, it is impossible to according to the VLAN Lai She Ge for allowing to pass through, but need to be counted according to TRILL The VLAN calculated cuts tree and carrys out Dui Ying She Ge（Needing to remove equipment upper port allows the VLAN and VLAN passed through broadcast The consistency check of the port is included in bitmap）.

For different VLAN, tree can be cut using different VLAN to realize the load balancing of broadcast traffic.Such as Fig. 3 Shown, the link of bold portion is VLAN1 broadcast domain, according to this link topology, the , She Ge VLAN1 on nickname-1 Bitmap includes user-side port P5 and up going port P1-P4；, She Ge VLAN1 bitmap only includes on nickname-g1 P1, P4 and Pn, user-side port P5 and up going port P1 is only included on nickname-4；On nickname-n, only include using Family side ports P5 and up going port P1.For bitmap She Ge principle same VLAN1s of the VLAN2 on each RB（The chain of dotted portion Road）, it is not repeated to describe herein.Need to broadcast in VLAN1 for the message received from nickname-1 user-side ports When, it is only necessary to broadcasted according to corresponding VLAN bitmap, the broadcasting packet can be along the chain of bold portion Road, the related RB of all VLAN1 are reached, and loop will not be produced.

Wherein, above-mentioned VLAN is cut tree and distribution tree progress beta pruning is obtained based on VLAN.Specifically, in Trill domains, For each VLAN, each distribution tree is required to be cut, to wipe out the branch of no potential recipient；In the VLAN Multicast/broadcast data can only forward on the branch that the distribution tree is not wiped out.

For TRILL mouths, VLAN broadcast domain scope specifically include above-mentioned VLAN cut tree by way of each RB on Land port, i.e. the VLAN cut set by way of each RB on exit port.

Further, in embodiments of the present invention, after MAC forward tables are issued to hardware, if RB receives two layers of unicast report Text, RB refusals carry out Trill encapsulation to two layers of unicast message（Trill encapsulation is not carried out to two layers of unicast message）, root According to two layers of unicast message target MAC (Media Access Control) address inquire about MAC address entries, and by two layers of unicast message in the form of native from inquiry To MAC address entries in exit port（egressport）Send.

With host1 in VLAN1 in scene shown in Fig. 3（MAC Address is mac1）To host7（MAC Address is mac7）Report Illustrated exemplified by literary forwarding process.Wherein, the MAC forwardings generated on nickname-1, nickname-g1, nickname-n Table can be with as shown in figure 3, specific as shown in following table：

MAC forward tables on table 4, nickname-1

VLAN	MAC	Egress port
			VLAN1	mac1	P5
VLAN1	mac4	Trunk（P1、P2、P3、P4）
			VLAN1	mac7	Trunk（P1、P2、P3、P4）
VLAN1	mac-v1	P1
			VLAN2	mac2	P6
VLAN2	mac5	Trunk（P1、P2、P3、P4）
			VLAN2	mac8	Trunk（P1、P2、P3、P4）
VLAN2	mac-v2	P1
			VLAN3	mac3	P7
VLAN3	mac6	Trunk（P1、P2、P3、P4）
			VLAN3	mac9	Trunk（P1、P2、P3、P4）
VLAN3	mac-v3	P1

MAC forward tables on table 5, nickname-g1

VLAN	MAC	Egress port
			VLAN1	mac1	P1
VLAN1	mac4	P4
			VLAN1	mac7	Pn
VLAN1	mac-v1	Local L3 are marked
			VLAN2	mac2	P1
VLAN2	mac5	P4
			VLAN2	mac8	Pn
VLAN2	mac-v2	Local L3 are marked
			VLAN3	mac3	P1
VLAN3	mac6	P2
			VLAN3	mac9	Pn
VLAN3	mac-v3	Local L3 are marked

MAC forward tables on table 6, nickname-n

VLAN	MAC	Egress port
			VLAN1	mac1	Trunk（P1、P2、P3、P4）
VLAN1	mac4	Trunk（P1、P2、P3、P4）
			VLAN1	mac7	P5
VLAN1	mac-v1	P1
			VLAN2	mac2	Trunk（P1、P2、P3、P4）
VLAN2	mac5	Trunk（P1、P2、P3、P4）
			VLAN2	mac8	P5
VLAN2	mac-v2	P1
			VLAN3	mac3	Trunk（P1、P2、P3、P4）
VLAN3	mac6	Trunk（P1、P2、P3、P4）

VLAN3	mac9	P5
			VLAN3	mac-v3	P1

As shown in figure 4, for a kind of schematic flow sheet of two layers of unicast message retransmission method provided in an embodiment of the present invention, can To comprise the following steps：

Step 401, host1 send message to nickname-1（Two layers of unicast message）, the target MAC (Media Access Control) address of the message is Mac7, affiliated VLAN are VLAN1.

After step 402, nickname-1 receive the message, according to the target MAC (Media Access Control) address of the message and affiliated VLAN Local MAC forward tables are inquired about, and nickname-g1 is forwarded the packet to according to the list item inquired.

Specifically, after nickname-1 receives the message of host1 transmissions, according to the target MAC (Media Access Control) address of the message （mac7）And affiliated VLAN（VLAN1）Local MAC forward tables are inquired about, hit nickname-1 local mac forward tables in Fig. 3 List item in middle dotted line circle, exit port is aggregation port in the list item（Trunk1（P1-P4））, nickname-1 calculated by hash Method from choose one of them, it is assumed that be port P1, and the message sent from P1.Wherein, nickname-1 is by the message When being forwarded from P1, Trill encapsulation will not be carried out to the message, but sent in the form of native by by the message.

After step 403, nickname-g1 receive the message, according to the target MAC (Media Access Control) address of the message and affiliated VLAN Local MAC forward tables are inquired about, and nickname-n is forwarded the packet to according to the list item inquired.

Specifically, after nickname-g1 receives the message of nickname-1 forwardings, according to the purpose MAC of the message Location（mac7）And affiliated VLAN（VLAN1）Local MAC forward tables are inquired about, nickname-g1 local macs in Fig. 3 is hit and turns The list item in the list item dotted line circle in dotted line circle is delivered, exit port is Pn in the list item, and the message is transferred from the port Go.Wherein, nickname-g1 does not also carry out Trill encapsulation to the message.

After step 404, nickname-n receive the message, according to the target MAC (Media Access Control) address of the message and affiliated VLAN Local MAC forward tables are inquired about, and host7 is forwarded the packet to according to the list item inquired.

Specifically, after nickname-n receives the message of nickname-g1 forwardings, according to the purpose MAC of the message Location（mac7）And affiliated VLAN（VLAN1）Local MAC forward tables are inquired about, nickname-n local macs in Fig. 3 is hit and forwards List item in table in dotted line circle, the exit port of the list item is P5, and the message is forwarded from the port.

Further, in embodiments of the present invention, in the network topology change of big double layer network（Such as find node device or Link failure etc. is abnormal）, cause big double layer network forward table and/or VLAN cuttings to be set after updating, node device can root Original MAC address forwarding-table item and/or corresponding VLAN broadcast domain scope are updated according to concrete condition.

Specifically, when the outgoing interface of the forwarding-table item in big double layer network forward table updates, node device can be with The outgoing interface of original MAC address forwarding-table item corresponding to renewal；

When the forwarding-table item in big double layer network forward table is deleted, node device can determine deleted forward table The destination node equipment of item and the local mac address for belonging to the destination node equipment, and then delete the destination node equipment Local mac address corresponding to original MAC address forwarding-table item；

When VLAN, which cuts tree, to be updated, it is corresponding that node device can cut tree Chong Xin She Ge according to the VLAN after renewal VLAN broadcast domain scope.

Still by taking the Trill technologies in data center as an example, the exception in the technical scheme provided the embodiment of the present invention Treatment mechanism is described in detail.

Scene one, Trill link failures

Referring to Fig. 5, it is assumed that the Trill links between nickname-1 and nickname-g1 break down, then with this The node device of Trill link direct correlation（That is nickname1 and nickname-g1）It needs to be determined that the net after the link failure Network topology, and the network topology is noticed to other node devices, it can such as be led to by ISIS protocol massages to other node devices Accuse the network topology after link failure.Each node device then needs to refresh itself according to the network topology after link failure Trill unicast routing tables and multicast table.

Wherein, for Trill unicast routing tables, the list item that only exit port changes, just MAC forwardings corresponding to refreshing The exit port of list item.

For example, the MAC forwarding-table items on nickname-1 to VLAN1/mac7 will be refreshed as table 8 by table 7：

Table 7

VLAN	MAC	Egress port
			VLAN1	Mac7	Trunk（P1、P2、P3、P4）

Table 8

VLAN	MAC	Egress port
			VLAN1	Mac7	Trunk（P2、P3、P4）

For Trill multicast tables, each node device needs to generate new broadcast tree according to the network topology after link failure, And refresh each VLAN broadcast domain scope again according to the new broadcast tree.Wherein, the broadcast domain scope after the refreshing includes Allow being cut out positioned at the VLAN in the user-side port and the node device that the VLAN passes through after link failure on each node device The big double layer network side interface on tree is cut, it is the beta pruning that tree is distributed based on the VLAN broadcast trees new to this that the VLAN, which cuts tree, Obtain.

Scene two, node device failure

Referring to Fig. 6, it is assumed that nickname-1 breaks down, then the node device associated with nickname-1 it needs to be determined that Network topology after the node device failure, and notice the network topology to other node devices.

Each node device needs to refresh the Trill unicast routing tables of itself according to the network topology after nickname-1 failures And multicast table.Wherein, therefore, will in the Trill unicast routing tables on other node devices because nickname-1 breaks down There is no the route for reaching nickname-1.

The control plane of each node device, which is deleted, in Trill networks belongs to the node to break down and sets on this node device It is standby（That is nickname-1）All MAC forwarding-table items（Such as the MAC of the MAC Address for the main frame hung under corresponding nickname-1 Forwarding-table item）.For Trill unicast routing tables, the list item that only exit port changes, just refresh corresponding to MAC forwarding-table items Exit port.

Correspondingly, similar with link failure scene, for Trill multicast tables, each node device is needed according to node Network topology after equipment fault generates new broadcast tree, and refreshes each VLAN broadcast domain again according to the new broadcast tree Scope.For example, in the scene shown in Fig. 6, on nickname-g1, VLAN1 broadcast domain scope will delete nickname-1 The port being connected with nickname-g1.

It should be appreciated that above-described embodiment turns by taking the Trill technologies in data center as an example to message provided by the invention Hair technical scheme is described in detail, when relative theory is applied to other big two layers of networking technologys（Such as Vxlan technologies, SPB skills Art）When in, corresponding description should be also adjusted.

For example, for access layer equipment and core layer equipment in big double layer network, in Trill technologies, Access Layer is set Standby and core layer equipment is RB；And in Vxlan technologies, access layer equipment VTEP（Vxlan Tunnel Endpoint, Vxlan endpoint of a tunnels）, and to core layer equipment not special definition；And in SPB technologies, access layer equipment BEB （Backbone Edge Bridge, backbone edge bridge）Equipment, nucleus equipment BCB（Backbone Core Bridge, it is key Core bridges）Equipment.

For big double layer network forward table, in Trill technologies, the big double layer network forward table is specially Trill routes Table；And in Vxlan technologies, the big double layer network forward table is specially the backbone network L3 routing tables in Vxlan；And in SPB skills In art, the big double layer network forward table is specially the B-MAC forward tables of the backbone network in SPBM.

For the control message synchronous for carrying out MAC Address, in Trill technologies, this is used to carry out MAC Address synchronization Control message be specially ESADI protocol massages；And in Vxlan technologies and SPB technologies, due to itself and undefined active MAC Address synchronization mechanism, therefore, the ESADI protocol massages in Trill technologies can be borrowed in Vxlan technologies and SPB technologies In, its specific implementation repeats no more herein.

Based on identical technical concept, the embodiment of the present invention additionally provides a kind of big two-layer network device, can apply to Above method embodiment.

As shown in fig. 7, for a kind of structural representation of big two-layer network device provided in an embodiment of the present invention, this is big two layers The network equipment can be applied to the big double layer network that multiple node devices are set up based on big double layer network agreement as node device, The big two-layer network device includes：

Address management module 71, for by one up to all node devices in the big double layer network broadcast tree, The local mac address and affiliated VLAN that the big two-layer network device is newly learnt are advertised to other node devices, described Local mac address includes the MAC Address learnt from the user network side of this equipment connection；

List item maintenance module 72, for receiving other node devices from the broadcast tree in the big two-layer network device After the first MAC Address and affiliated first VLAN that other described node devices of notice newly learn, according to other described sections The big double layer network mark of point device, inquiry are in advance based on the big double layer network forward table of the big double layer network protocol generation, It is determined that outgoing interface corresponding to other described node devices, and then generate original MAC address forwarding corresponding to first MAC Address List item, wherein, the VLAN in the original MAC address forwarding-table item is the first VLAN, target MAC (Media Access Control) address is described first MAC Address, outgoing interface are outgoing interface corresponding to other described node devices；

Packet forwarding module 73, for after the big two-layer network device receives unicast data message, according to described The target MAC (Media Access Control) address of unicast data message and affiliated VLAN, matched, and led in the original MAC address forwarding-table item The outgoing interface of overmatching list item directly forwards the unicast data message and without big double layer network protocol encapsulation.

Wherein, the list item maintenance module 72 can be specifically used for, and be accomplished by the following way and set according to other described nodes Standby big double layer network mark, inquiry are in advance based on the big double layer network forward table of the big double layer network protocol generation, it is determined that Outgoing interface corresponding to other described node devices：

Destination node equipment is identified as with the big double layer network of other node devices, forwarded from the big double layer network In table, forwarding-table item corresponding to the big double layer network mark of other node devices is found；

If only including an outgoing interface in the forwarding-table item found, the outgoing interface that will be found, as it is described other Outgoing interface corresponding to node device；

If the forwarding-table item found includes more than two outgoing interfaces, more than two outgoing interfaces are formed one Individual aggregation port, and using the aggregation port as outgoing interface corresponding to other described node devices.

Wherein, the packet forwarding module 73 can be specifically used for, and be accomplished by the following way and go out to connect by matching list item Mouthful directly forward the unicast data message and without big double layer network protocol encapsulation：

If the outgoing interface for matching list item is aggregation port, pass through a member of aggregation port described in HASH algorithms selections Port, and by selected member port, directly forward the unicast data message and sealed without big double layer network agreement Dress.

Wherein, the address management module 71 can be additionally used in, and be set by the broadcast, the local mac of aging will occur Location and affiliated VLAN are advertised to other node devices；

The list item maintenance module 72 can be additionally used in, from it is described broadcast tree on receive other node devices notice described in its The second MAC Address of aging and affiliated VLAN occur at his node device, and from the original MAC address forwarding-table item, Delete second MAC Address and the forwarding-table item corresponding to affiliated VLAN.

Wherein, the big two-layer network device is directed to the combination of itself different interface and VLAN, point not Pei be corresponding to Ge mono- MAC Address pond, for being maintained in the local mac address for belonging to corresponding VLAN learnt on the corresponding interface；

The address management module 71 can be specifically used for, when local mac address is arrived in new study, if corresponding MAC Address pond In do not include this it is new learn local mac address when, then the new local mac address learnt is increased into MAC Address pond In, and, after one section of preset time that aging occurs for local mac address, by the local mac address of aging from corresponding MAC Deleted in pond location；When the local mac address newly learnt is added in MAC Address pond, set, will newly learnt by the broadcast To local mac address and affiliated VLAN be advertised to other node devices；And delete aging in MAC Address pond During local mac address, set by the broadcast, the local mac address of aging and affiliated VLAN are advertised to other nodes and set It is standby.

Wherein, the packet forwarding module 71 can be additionally used in,

The She Ge VLAN broadcast domain scope Shang node device, wherein, the broadcast domain scope is included on the node device Allow the big double layer network side joint cut positioned at the VLAN on tree in the user-side port and the node device that the VLAN passes through Mouthful, the cutting tree is obtained based on the VLAN beta prunings for being distributed tree to the broadcast tree；

After receiving data message to be broadcast, the affiliated VLAN of data message to be broadcast is determined, and in the wide of the VLAN Broadcast on the port of other in addition to receiving port included by the scope of domain, directly forward the data message to be broadcast and without big Double layer network protocol encapsulation, wherein, the data message to be broadcast includes unknown unicast message, broadcasting data messages and unknown group Report text.

Wherein, the list item maintenance module 72 can be additionally used in, and in the network topology change of the big double layer network, cause big After double layer network forward table updates, if the outgoing interface of the forwarding-table item in big double layer network forward table updates, update The outgoing interface of corresponding original MAC address forwarding-table item；If the forwarding-table item in big double layer network forward table is deleted, it is determined that quilt The destination node equipment of the forwarding-table item of deletion and the local mac address for belonging to the destination node equipment, and then delete and be somebody's turn to do Original MAC address forwarding-table item corresponding to the local mac address of destination node equipment；

The packet forwarding module is additionally operable to, and in the network topology change of the big double layer network, causes VLAN to cut tree After updating, tree is cut according to the VLAN after renewal, Chong Xin She Ge correspond to VLAN broadcast domain scope.

Through the above description of the embodiments, those skilled in the art can be understood that the present invention can be by Software adds the mode of required general hardware platform to realize, naturally it is also possible to which by hardware, but the former is more in many cases Good embodiment.Based on such understanding, what technical scheme substantially contributed to prior art in other words Part can be embodied in the form of software product, and the computer software product is stored in a storage medium, if including It is dry to instruct to cause a station terminal equipment（Can be mobile phone, personal computer, server, or network equipment etc.）Perform sheet Invent the method described in each embodiment.

Described above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should Depending on protection scope of the present invention.

Claims (14)

1. a kind of message forwarding method, the big double layer network set up applied to multiple node devices based on big double layer network agreement In, it is characterised in that this method includes：

Each node device by one up to all node devices in the big double layer network broadcast tree, learn new Local mac address and the affiliated VLAN of the local mac address are advertised to other node devices, and the local mac address includes The MAC Address learnt from the user network side of this equipment connection；

The node device from it is described broadcast tree on receive other node devices notice described in other node devices newly study to The first MAC Address and affiliated first VLAN after, identified according to the big double layer network of other node devices, inquiry is advance Based on the big double layer network forward table of the big double layer network protocol generation, it is determined that going out to connect corresponding to other described node devices Mouthful, and then original MAC address forwarding-table item corresponding to first MAC Address is generated, wherein, the original MAC address forwarding VLAN in list item is the first VLAN, target MAC (Media Access Control) address is first MAC Address, outgoing interface is other described nodes Outgoing interface corresponding to equipment；

After the node device receives unicast data message, according to the target MAC (Media Access Control) address of the unicast data message and affiliated VLAN, matched in the original MAC address forwarding-table item, and the outgoing interface by matching list item directly forwards the list Multicast data message and without big double layer network protocol encapsulation.

2. the method as described in claim 1, it is characterised in that the big double layer network mark of other node devices described in the basis Know, inquiry is in advance based on the big double layer network forward table of the big double layer network protocol generation, it is determined that other described node devices Corresponding outgoing interface, including：

Destination node equipment is identified as with the big double layer network of other node devices, from the big double layer network forward table In, find forwarding-table item corresponding to the big double layer network mark of other node devices；

If only include an outgoing interface, the outgoing interface that will be found, as other described nodes in the forwarding-table item found Outgoing interface corresponding to equipment；

If the forwarding-table item found includes more than two outgoing interfaces, more than two outgoing interfaces are formed one and gathered Port is closed, and using the aggregation port as outgoing interface corresponding to other described node devices.

3. method as claimed in claim 2, it is characterised in that the outgoing interface by matching list item directly forwards the list Multicast data message and without big double layer network protocol encapsulation, including：

If the outgoing interface for matching list item is aggregation port, pass through a member end of aggregation port described in HASH algorithms selections Mouthful, and by selected member port, directly forward the unicast data message and without big double layer network protocol encapsulation.

4. the method as described in claim 1, it is characterised in that also include：

Each node device is also set by the broadcast, and the local mac address that aging occurs and affiliated VLAN are advertised into it His node device；

Node device receives from the broadcast tree occurs aging described in the notice of other node devices at other node devices Second MAC Address and affiliated VLAN, and from the original MAC address forwarding-table item, delete second MAC Address and Forwarding-table item corresponding to affiliated VLAN.

5. the method as described in claim 1, it is characterised in that each node device is for itself different interface and VLAN Combination, is respectively configured MAC Address pond corresponding to one, for be maintained in learn on the corresponding interface belong to correspondence VLAN Local mac address；

Each node device new learns if not including this in correspondence MAC Address pond when local mac address arrive in new study During local mac address, then the new local mac address learnt is increased in MAC Address pond, and, sent out in local mac address After one section of preset time of raw aging, the local mac address of aging is deleted from corresponding MAC Address pond；

Each node device, when the local mac address newly learnt is further added in MAC Address pond, by described wide Tree is broadcast, the new local mac address learnt and affiliated VLAN are advertised to other node devices；And in MAC Address pond When deleting the local mac address of aging, set by the broadcast, the local mac address of aging and affiliated VLAN are noticed Give other node devices.

6. the method as described in claim 1, it is characterised in that this method also includes：

VLAN broadcast domain scope is set on node device, wherein, the broadcast domain scope includes allowing on the node device The big double layer network side interface cut positioned at the VLAN on tree in the user-side port and the node device that the VLAN passes through, institute Cutting tree is stated to obtain distribution tree progress beta pruning based on VLAN；

After node device receives data message to be broadcast, the affiliated VLAN of data message to be broadcast is determined, and in the VLAN Broadcast domain scope included by the port of other in addition to receiving port on, directly forward the data message to be broadcast without entering The big double layer network protocol encapsulation of row, wherein, the data message to be broadcast includes unknown unicast message, broadcasting data messages and not Know multicast message.

7. method as claimed in claim 6, it is characterised in that in the network topology change of the big double layer network, cause big After double layer network forward table and/or VLAN cutting trees update, this method also includes：

It is former corresponding to renewal when the outgoing interface of forwarding-table item of the node device in big double layer network forward table updates The outgoing interface of beginning mac address forwarding table item；

When forwarding-table item of the node device in big double layer network forward table is deleted, it is determined that deleted forwarding-table item Destination node equipment and the local mac address for belonging to the destination node equipment, and then delete the sheet of the destination node equipment Original MAC address forwarding-table item corresponding to ground MAC Address；

The node device cuts tree according to the VLAN after renewal, reset correspondingly after VLAN cutting trees update VLAN broadcast domain scope.

8. a kind of big two-layer network device, it is applied to multiple node devices as node device and is set up based on big double layer network agreement Big double layer network, it is characterised in that the big two-layer network device includes：

Address management module, for by one up to all node devices in the big double layer network broadcast tree, will described in The local mac address and the affiliated VLAN of the local mac address that big two-layer network device newly learns are advertised to other nodes Equipment, the local mac address include the MAC Address learnt from the user network side of this equipment connection；

List item maintenance module, for receiving the notice of other node devices from the broadcast tree in the big two-layer network device After the first MAC Address and affiliated first VLAN that other described node devices newly learn, according to other described node devices Big double layer network mark, inquiry is in advance based on the big double layer network forward table of the big double layer network protocol generation, determines institute Outgoing interface corresponding to other node devices is stated, and then generates original MAC address forwarding-table item corresponding to first MAC Address, Wherein, the VLAN in the original MAC address forwarding-table item be the first VLAN, target MAC (Media Access Control) address be the first MAC Location, outgoing interface are outgoing interface corresponding to other described node devices；

Packet forwarding module, for after the big two-layer network device receives unicast data message, according to the unicast number Target MAC (Media Access Control) address and affiliated VLAN according to message, are matched, and pass through matching in the original MAC address forwarding-table item The outgoing interface of list item directly forwards the unicast data message and without big double layer network protocol encapsulation.

9. big two-layer network device as claimed in claim 8, it is characterised in that the list item maintenance module is specifically used for, and leads to The big double layer network mark that in the following manner realizes other node devices according to is crossed, inquiry is in advance based on the big double layer network The big double layer network forward table of protocol generation, it is determined that outgoing interface corresponding to other described node devices：

Destination node equipment is identified as with the big double layer network of other node devices, from the big double layer network forward table In, find forwarding-table item corresponding to the big double layer network mark of other node devices；

If only include an outgoing interface, the outgoing interface that will be found, as other described nodes in the forwarding-table item found Outgoing interface corresponding to equipment；

If the forwarding-table item found includes more than two outgoing interfaces, more than two outgoing interfaces are formed one and gathered Port is closed, and using the aggregation port as outgoing interface corresponding to other described node devices.

10. big two-layer network device as claimed in claim 9, it is characterised in that the packet forwarding module is specifically used for, and leads to Cross in the following manner realization and directly forward the unicast data message by matching the outgoing interface of list item and without big double layer network Protocol encapsulation：

If the outgoing interface for matching list item is aggregation port, pass through a member end of aggregation port described in HASH algorithms selections Mouthful, and by selected member port, directly forward the unicast data message and without big double layer network protocol encapsulation.

11. big two-layer network device as claimed in claim 8, it is characterised in that

The address management module is additionally operable to, and is set by the broadcast, by the local mac address that aging occurs and affiliated VLAN It is advertised to other node devices；

The list item maintenance module is additionally operable to, and other nodes described in receiving the notice of other node devices are set from the broadcast and are set Second MAC Address of aging and affiliated VLAN occur for standby place, and from the original MAC address forwarding-table item, described in deletion Second MAC Address and the forwarding-table item corresponding to affiliated VLAN.

12. big two-layer network device as claimed in claim 8, it is characterised in that the big two-layer network device is directed to itself not The combination of same interface and VLAN, is respectively configured MAC Address pond corresponding to one, for being maintained in what is learnt on the corresponding interface Belong to corresponding VLAN local mac address；

The address management module is specifically used for, when local mac address is arrived in new study, if not included in corresponding MAC Address pond Have this it is new learn local mac address when, then the new local mac address learnt is increased in MAC Address pond, and, After one section of preset time that aging occurs for local mac address, the local mac address of aging is deleted from corresponding MAC Address pond Remove；When the local mac address newly learnt is added in MAC Address pond, set by the broadcast, the local that new will learn MAC Address and affiliated VLAN are advertised to other node devices；And the local mac of aging is deleted in MAC Address pond During location, set by the broadcast, the local mac address of aging and affiliated VLAN are advertised to other node devices.

13. big two-layer network device as claimed in claim 8, it is characterised in that the packet forwarding module is additionally operable to,

VLAN broadcast domain scope is set on node device, wherein, the broadcast domain scope includes allowing on the node device The big double layer network side interface cut positioned at the VLAN on tree in the user-side port and the node device that the VLAN passes through, institute Cutting tree is stated to obtain distribution tree progress beta pruning based on VLAN；

After receiving data message to be broadcast, the affiliated VLAN of data message to be broadcast is determined, and in the broadcast domain of the VLAN On the port of other in addition to receiving port included by scope, the data message to be broadcast is directly forwarded and without big two layers Procotol encapsulates, wherein, the data message to be broadcast includes unknown unicast message, broadcasting data messages and unknown multicast report Text.

14. big two-layer network device as claimed in claim 13, it is characterised in that

The list item maintenance module is additionally operable to, and in the network topology change of the big double layer network, causes big double layer network to forward After table updates, if the outgoing interface of the forwarding-table item in big double layer network forward table updates, original MAC corresponding to renewal The outgoing interface of address repeat listing；If the forwarding-table item in big double layer network forward table is deleted, it is determined that deleted forward table The destination node equipment of item and the local mac address for belonging to the destination node equipment, and then delete the destination node equipment Local mac address corresponding to original MAC address forwarding-table item；

The packet forwarding module is additionally operable to, and in the network topology change of the big double layer network, causes VLAN to cut tree After renewal, tree is cut according to the VLAN after renewal, resets corresponding VLAN broadcast domain scope.