CN1581828A - Method for realizing dual attachment of client boundary device - Google Patents
Method for realizing dual attachment of client boundary device Download PDFInfo
- Publication number
- CN1581828A CN1581828A CN 03154137 CN03154137A CN1581828A CN 1581828 A CN1581828 A CN 1581828A CN 03154137 CN03154137 CN 03154137 CN 03154137 A CN03154137 A CN 03154137A CN 1581828 A CN1581828 A CN 1581828A
- Authority
- CN
- China
- Prior art keywords
- route
- incidence relation
- virtual
- routing table
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The method includes steps: in virtual route table of device of service provider, based on first association relation formed from data structure of storing original route information, and second reference relation formed from data structure of storing route reference relation, third association relation is formed from data structure of reference relation through all routes introduced in same virtual forwarding instance and going to same destination; when re-optimizing routes caused by route information being added or deleted is needed, global route selection is carried out among different virtual route tables based on first, second and third relations.
Description
Technical field
The present invention relates to virtual private network technology, the dual-homed implementation method of particularly a kind of client boundary equipment.
Background technology
In the dual-homed networking of client boundary equipment (be called for short CE), because a CE connects two server's edge devices (being called for short PE) simultaneously, it is dual-homed that this network construction form becomes CE.Consult Fig. 1, client boundary equipment CE1 is connected with PE2 with server's edge device PE1 respectively, therefore a VPN route (for example 10.163.1.1/24) of its inside server's edge device PE1 and the PE2 that will pass through respectively to be connected releases, and other end server edge device PE3 will receive this two routes to same destination from server's edge device PE1 and PE2 respectively.If these two routes are placed in the different routing tables, do not select excellent mechanism because present realization is carried out route between different routing tables, so the Border Gateway Protocol (bgp protocol) that server's edge device PE3 passes through these two routes between the PE3-CE2 has been issued client boundary equipment CE2.Because their destination, subnet mask and next jumping are all identical, so client boundary equipment CE2 thinks a route, the route that transmits after only keeping.Yet after a route chain rupture wherein, remote service person's edge device PE1 is by bgp protocol notice local terminal server edge device PE3, local terminal server edge device PE3 just sends unreachable message to client boundary equipment CE2 then, causes client boundary equipment CE2 with that route deletion.If the route-distinguisher of two routes that far-end transmits (being called for short RD) difference, during they are placed on and (are called for short VRIB) in two different virtual routing tables, therefore wherein a route deleted after, can not influence the identical destination route that is arranged in another VRIB, so can not send route to client boundary equipment CE2 once more yet, though thereby cause on server's edge device PE3, also having forward-path, but on client boundary equipment CE2, there is not corresponding route, thereby the user of CE back can't pass through the routing device of the access carrier network that another strip label switching path (be called for short LSP) visit far-end safeguarded by user oneself and the set (being called for short site) of main frame, makes and wants originally that the original intention by backup LSP increase reliability can't realize.
For the problem in the above-mentioned networking, have three kinds to evade scheme:
Scheme one, two server's edge device PE1 that client boundary equipment CE1 is linked to each other are configured to identical RD with virtual routing transmitting example (being called for short VRF) on the PE2, so just can continue to use former route refresh flow process, from being placed on the same VRIB at server's edge device PE3 that server's edge device PE1 and PE2 transmit to same destination route, and existing mechanism can be selected optimum route sending-down client boundary equipment CE2 in VRIB, can select route stand-by automatically after the route in being elected to is unreachable.
But the prerequisite that this scheme requires: the VRF on the far-end PE is configured to identical RD, may not satisfy in actual networking.Because remote equipment may not be the equipment with manufacturer, may come into the market at different time, and operational excellence.If change RD, then need to issue again all routes among this VRF, the fluctuation that brings network, the operator may be difficult to accept.
Scheme two, the configuring static route is pointed to server's edge device PE3 on client boundary equipment CE2, so just can not be subjected to the influence that route refreshes on server's edge device PE3.
But static routing also has its limitation.It can not when particularly the far-end destination is unreachable, also be sent to public network with flow according to the topological real-time change on the current network, brings unnecessary impact to network.In addition, these static routing need be disposed personnel oneself and be safeguarded, particularly work as number of routes more for a long time, and the workload of maintenance is very big, and makes mistakes easily.Therefore, user's CE equipment might not accept to rely on fully the scheme of static routing.
Scheme three, between server's edge device PE3 and client boundary equipment CE2 operation RIP agreement.Because being the mode of regularly issuing all routes, the RIP agreement circulates a notice of the route variation.Even server's edge device PE3 does not issue CE with another route stand-by, but through after a while after (as 30 seconds), the RIP consultation again with current all can reach route and all issue client boundary equipment CE2.But because the RIP agreement is comparatively outmoded, the mode of particularly the sort of all routes of periodic broadcasting can be brought certain impact to network more for a long time in route, so user's CE equipment might be unwilling to move RIP between PE-CE.
Three kinds of top schemes all are in the scheme of evading that has under certain assumed condition, are not the real solution to this problem.And, except the CE above-mentioned problem that route can not automated back-up when dual-homed also, also have some other problem: for example the order of the maximum number of routes (maxroute) in the restriction routing table of VRF can only limit local route and can not limit the far-end route; If the derivation route target (export route target) of two routes in routing table is configured to not simultaneously, some VRF will can not receive problems such as route.The all same reason of all these problems causes, promptly present multirouting list structure improves inadequately and to cause.
Present multirouting list structure is in order to satisfy the MPLS/VPN modelling, and it is placed on the route that different site come in the different routing tables, uses the RD unique identification.Intercommunication between the different site realizes the mutual of route by the mutual coupling of route target (Route target).Therefore just exist a route among the VRIB can import a plurality of different VRF, and VRF can receive the situation of the route among a plurality of VRIB, the cross reference of a kind of like this m*n has embodied the flexibility of the height of MPLS/VPN, but has also increased the complexity that realizes simultaneously.
Present implementation mainly is in order to solve memory problem.Route among site only is present among the VRIB of its RD correspondence, realizes transmitting copy with the route issue by revising refresh flow.Internal memory has not only been saved in such realization, and the inconsistency between the data brought of the multiple copy that has saved route, from this point a kind of outstanding solution of can yet be regarded as, it has successfully solved the problem that a VRIB in the cross reference imports a plurality of VRF.
But for the another kind of problem in the cross reference, promptly VRF receives the route of route, particularly same destination among a plurality of VRIB, and this scheme is failed fine solution.Mainly be since between a plurality of VRIB without any association, do not know each other.Therefore, the route between the different VRIB can't compare, select excellent and mutual backup, and the route among VRIB changes the VRIB that can not notify other.Just because of having cut off such connecting each other, caused occurring above-described problem.
Though, can temporarily solve by the method for patch installing for the problem of present appearance, will make the code of route become increasingly complex like this, becoming is difficult to safeguard, has more increased the possibility that goes wrong, and has also lost the ability that adapts to new demand that further expands simultaneously.
Summary of the invention
The object of the present invention is to provide the dual-homed implementation method of a kind of client boundary equipment, to solve in server's edge device because of the unconnected overall situation comparison that can't carry out route between the different virtual routing table, to select the problem of excellent and mutual backup.
Technical scheme of the present invention:
The dual-homed implementation method of a kind of client boundary equipment, described client boundary equipment links to each other with two server's edge devices respectively, on server's edge device, have a plurality of Virtual Switch Instance and with each Virtual Switch Instance corresponding virtual routing table, and the virtual routing table of at least one far-end, routing iinformation to identical destination in same virtual routing table forms first incidence relation, and each routing iinformation can import the formation of all references relation and corresponding second incidence relation of this route of other Virtual Switch Instance; Described method comprises step:
Can import same Virtual Switch Instance and, set up the 3rd incidence relation by described adduction relationship to all routes of identical destination; And
Cause that adding routing iinformation and deletion routing iinformation route selects when excellent again, between the different virtual routing table, carry out overall Route Selection according to described first, second incidence relation and the 3rd incidence relation.
Wherein:
Described first, second is respectively a two-way chained list with the 3rd incidence relation.
Described adding routing iinformation comprises step:
After A, native client person edge device are received VPN route from the client person border of far-end, seek the virtual routing table of corresponding far-end in this locality according to the route-distinguisher under the route in the protocol massages, if find then carry out step B, otherwise with this route-distinguisher is that key value (KEY) is created the virtual routing table of new far-end, and continues step B;
B, newly distribute a data structure of depositing route, and fill with the routing iinformation of newly receiving;
C, this data structure of depositing route is inserted in first incidence relation of described far-end routing table, selected excellent rule in this virtual routing table, to carry out route according to route and select excellent;
D, according to derivation route target (export route-tagret) attribute of described route, importing route target (import route-tagte) attribute that mates local all Virtual Switch Instance successively, if can mate, then distribute one to deposit the data structure of route adduction relationship and be inserted in second incidence relation of this route of far-end routing table for Virtual Switch Instance;
E, the head node of in the Virtual Switch Instance corresponding virtual routing table that can import, searching first incidence relation identical with the purpose IP address and the mask of newly-increased route with it;
If do not find, then in current Virtual Switch Instance corresponding virtual routing table, create a new head node, the data structure of depositing the route adduction relationship of distributing among the step D is inserted in corresponding the 3rd incidence relation, selected excellent rule between this virtual routing table, to carry out route according to route and select excellent;
If find, then directly the data structure of depositing the route adduction relationship of distributing among the step D is inserted in corresponding the 3rd incidence relation, select excellent rule between this virtual routing table, to carry out route according to route and select excellent.
Described deletion routing iinformation comprises step:
Route-distinguisher (route-distinguish) attribute that a, basis wait to delete route finds the virtual routing table of depositing this route, and finds the actual data structure of waiting to delete route of depositing;
B, judge whether wait to delete route is current optimum route, if not then carrying out step c, otherwise it is excellent to trigger selecting again of route in first incidence relation, realizes the backup and the switching of route, continues step c;
C, handle and corresponding second incidence relation of this route successively, and trigger node in the 3rd incidence relation and between different virtual routing tables, select excellently again, realize the backup and the switching of route.
The present invention has following beneficial effect:
(1) problems such as can not realizing route backup can not be issued and transmit to the present invention's reached at the route that need not preset the EBGP that brings when any assumed condition just can thoroughly solve the CE dual-homing networking.
(2) the present invention can support such as: 1) by max route control far-end route; 2) can dispose the functions such as export route target that are arranged in same VRIB route arbitrarily
(3) a virtual routing table can be imported a plurality of Virtual Switch Instance and Virtual Switch Instance can accept the relation of m*n of this complexity of route of a plurality of virtual routing tables and combine, for the route target (route target) of route and Virtual Switch Instance without any restriction, rank that can control fine size to a route, this be former scheme can not support.The flexibility of height has been arranged just, made it support aspect the new demand very big autgmentability is arranged.
Description of drawings
Fig. 1 is the dual-homed networking schematic diagram of client boundary;
Fig. 2 is the schematic diagram of the interior routing relation of virtual routing table in server's edge device of the present invention.
Fig. 3 adds the flow chart of route for the present invention;
Fig. 4 deletes the flow chart of route for the present invention.
Embodiment
The present invention imitates between the route that is associated in different VRIB between the route of the identical destination in the same virtual routing table (be called for short VRIB) and sets up contact, uses that the mutual back mechanism between the identical route solves the backup problem between the route of identical destination between the different VRIB among the existing same VRIB.Such contact need meet the following conditions:
(1) route of different VRIB can mutual perception, can resemble to use all existing comparison rule select optimum route same VRIB.After changing or better route occurs, optimum route can reselect and issue again list item automatically.
(2) any route among the VRIB should allow to import arbitrarily in the Virtual Switch Instance (be called for short VRF) of appointment, and any one VRF also should allow to import any route among any VRIB, the relation of these two aspects should without any mutual restriction.And for a VRF, its only is concerned about the information that imports route wherein and is indifferent to these routes and is arranged in which VRIB.
(3) though VRF can import the route among a plurality of VRIB, but when the VPN(Virtual Private Network) example of Border Gateway Protocol (BGP) issue VPN route, can only issue the VPN route of oneself and other private network can not be issued in the lump, therefore, it must distinguish those is the routes that import, and which is the route that oneself produces.
(4) in order to keep the stable of system, this change is limited within the scope of routing management as far as possible, reduces the influence to each Routing Protocol; Use original very ripe flow process as far as possible.
Join shown in Figure 2, Virtual Switch Instance VRF1 and VRF2 represent a Virtual Switch Instance on server's edge device (being called for short PE) respectively, it has represented the routing property by the set (being called for short site) of the routing device and the main frame of the access carrier network of user oneself maintenance that PE hangs down, comprises that route-distinguisher (route-distinguish, route-target), maximum route limit, whether can visit a series of attributes such as Internet.Each VRF is also related with a virtual routing table (VRIB) simultaneously, what deposit among the VRIB is exactly the VPN route that comprises among the local site, these routes can be diffused into remote equipment by bgp protocol and get on, so that the existence that the user of far-end can these routes of perception.But not all VRIB is associated with VRF, and the VPN route of receiving from far-end PE just need be placed into the brand-new far-end VRIB so that import in local VRF where necessary.These VPN routes form with radix tree in VRIB exists, and the route to identical destination (same subnet and mask-length) that is produced by different agreement all hangs in the doubly linked list with the head node rt_head that deposits the route chained list in the system.Write down the total attribute of these routes among the rt_head, for example IP address, mask etc.Node rt_entry is exactly a data structure of really depositing routing iinformation, and all node rt_entry constitute a doubly linked list to form first incidence relation among the VRIB.In order to represent the adduction relationship of route inside, on node rt_entry, haul out the doubly linked list that a node rtbit who deposits the data structure of route adduction relationship in by system forms again, form second incidence relation.The corresponding VRF that can import this route of each node in the chained list has write down route and has been quoted by which agreement among this VRF in the node.Data sharing and the problem quoted mutually when quoting chained list and solved that a route can import a plurality of VRF simultaneously by such one.
Can import same VRIB and, constitute doubly linked list, form the 3rd incidence relation, and make among the different VRIB that import among the VRF route interrelated, consult the chained list shown in the dotted line among the figure by node rtbit to all routes of identical destination.After the chained list of such all rtbit of series connection has been arranged, begin chain need be placed on the head node rt_head among the VRIB of the VRF correspondence that need to import, and in head node rt_head, expand an activity route pointer active_vrf and point to activation route in this chain, so that itself and original pointer active are made a distinction.
Because the route of the identical destination among the same VRIB is all gone here and there in a doubly linked list, any newly-increased node rt_entry or delete a node rt_entry, can in existing node rt_entry, select again excellent, thereby play the effect that route backup switches.Because the 3rd incidence relation also all is together in series the route of the identical destination among the different VRIB, therefore the route of the identical destination among the different VRIB each other can mutual perception, make and in a VRIB, increase newly or delete a route and can not only select excellent in VRIB inside, and can between VRIB, carry out the overall situation and select excellent, when not had route to this destination can reach in a VRIB, system can initiatively switch on the route of another VRIB.
Consult Fig. 2, the VRIB of two Virtual Switch Instance VRF1 and a VRF2 and a far-end is arranged in this locality, and have all between them that import to close be adding and the delete procedure that example illustrates route with PE.By second incidence relation, route among the VRIB of far-end can import among Virtual Switch Instance VRF1 and the VRF2, route as them is equally used, node rtbit in second incidence relation just likes the reflection of node rt_entry structure in different VRF of first incidence relation, just can find the rt_entry that really deposits routing iinformation by this reflection.Interrelated problem among the different VRIB that the 3rd incidence relation is enough realized importing among the VRF during route.
Consult Fig. 3, the concrete steps that PE adds route are as follows:
Step 10: after receiving VPN route from far-end PE, seek VRIB in this locality according to the route-distinguisher under the route in the protocol massages (route-distinguish).If find then carry out step 20, if do not find, be that key value (KEY) is created a new far-end VRIB just with this route-distinguisher, continue step 20.
Step 20: newly distribute a node rt_entry who deposits routing iinformation, and fill, comprise next jumping, label, metric etc. with the routing iinformation of newly receiving.
Step 30: this node rt_entry is inserted in VRIB first incidence relation that finds, promptly be suspended to on head node rt_head and the formed doubly linked list of node rt_entry.Can cause that selecting again of route in this chained list is excellent this moment.If optimum route changes, need notice MBGP agreement to issue these VPN routes again.
Step 40: according to derivation route target (export route-tagret) attribute of this route, mate importing route target (import route-tagte) attribute of local VRF successively, the derivation route target community of route obtains in protocol massages.If can mate, show that then this route can import this VRF, should be node rtbit who deposits the data structure of adduction relationship of this VRF distribution this moment in far-end VRIB, and it is suspended on the node rt_entry.
Step 50: with the purpose IP of newly-increased route and mask is sought first incidence relation in the VRIB of the VRF correspondence that can import head node rt_head (this is the radix searching algorithm of standard).If do not find, the route of also not receiving this destination among this VRF is described, then create a new head node rt_head.Then, node rt_entry in the step 40 being gone up the node rtbit that distributes for this VRF is inserted on the chained list of the 3rd incidence relation on this head node rt_head.It is excellent that may trigger selecting again of node rtbit in the chained list this moment.If optimum node rtbit changes, then issue the optimum route of newly selecting again by each Routing Protocol among this VRF.
Consult Fig. 4, the concrete steps of deletion route are as follows:
Step 100: route-distinguisher (route-distinguish) attribute according to route finds the VRIB routing table of depositing this route, uses the radix algorithm of standard to find the actual node rt_entry that waits to delete route that deposits then in first incidence relation.
Step 200: if this route is an optimum route at that time, it is excellent to trigger selecting again of route in the node rt_entry chained list, realizes the backup and the switching of route.If can select new optimum route, then the optimum route of newly selecting by the MBGP protocol issuance is not cancelled message otherwise notify MBGP to issue.
Step 300: handle vertical node rtbit that this node rt_entry hangs down second incidence relation successively, use the rule in the step 200, the node rtbit in the 3rd incidence relation is selected again excellent, realize the backup and the switching of route.Because all node rtbit that can import the route of this VRF go here and there on such doubly linked list (the 3rd incidence relation), then can between VRIB, select optimum route, avoided route between the original VRIB problem that can't be mutually related.When having Routing Protocol to quote this route among this VRF, the node rtbit deletion of correspondence when no longer quoting this route, can not deleted whole node rt_entry all VRF in the lump yet at last.
After setting up the 3rd incidence relation, can realize four above-mentioned demands:
1) chained list formed of node rtbit all is a doubly linked list with the chained list of original node rt_entry composition, and, therefore string all multinode rtbit are together selected the excellent satisfactory best route that still can obtain according to original rule by the node rt_entry that node rtbit can directly obtain correspondence.All adduction relationships that can import the identical destination route of VRF have all been gone here and there on a chained list simultaneously, therefore can select excellent again according to original flow process when wherein any route changes.So just satisfied first demand.
2) data structure of depositing adduction relationship is to deposit the reflection of data structure in different VRF of routing relation, difference of depositing the data structure of routing relation is deposited between the data structure of adduction relationship without any relation, so after a route can import a new VRF, only need newly-increased node rtbit, can not have any impact to existing node; Otherwise when a route no longer needs to import certain VRF, also only need the node rtbit of this VRF correspondence is got final product from the chained list excision.Make a route to import among the different VRF arbitrarily just because of the loose coupling between the node rtbit structure by the different data structures of depositing adduction relationship, deposit among which VRIB and the data structure of depositing routing relation that need not consider this route is actual, so also satisfied the second demand.
3) each VRF is mentioned in the front all a related with it VRIB, has wherein deposited local VPN route.These VPN routes associate by the chained list in first incidence relation, also only need the route on this chained list of diffusion when far-end PE spreads.And the route that other VRIB imports this VRF associates (the data structure structure of depositing adduction relationship that wherein also comprises local VPN route certainly) by the chained list in second incidence relation, when spreading route, local CE just needs the route on this chained list of diffusion again, so just the route on the both direction is diffused through and use different chained lists to make a distinction, so just satisfied the 3rd demand.
4) to chaining, tear chain open and select excellent operation all to finish by routing management, irrelevant with Routing Protocol.Therefore the chained list that increases such one the 3rd incidence relation can not have influence on Routing Protocol.Unique interface relevant with Routing Protocol is when notifying each agreement to issue route again in the flash flow process, need obtain the route of originally having issued.This search procedure is originally that chained list in first incidence relation realizes by traveling through, and just needs now to realize by the doubly linked list of the 3rd incidence relation pointed of active_vrf pointer among the traversal rt_head.In order to reduce from now on the influence that expansion once more brings, can in routing management, encapsulate a function and call for each agreement.And when obtaining up-to-date issue route, because therefore agreement and be indifferent to this route and whether be present among this VRIB only needs to replace pointer active with pointer active_vrf and get final product.This shows that the impact that this scheme is brought the route agreement is little, can satisfy the 4th demand.
Owing to set up the 3rd incidence relation, other some problems are naturally being solved.For example,
(1) originally in order to solve CE backuping each other between forwarding-table item when dual-homed, revises at forwarding module.But such modification is not too reasonable, because optimum route has the selection criterion of oneself, sometimes this criterion is also very complicated and new variation might occur, and forwarding module is not noted this, and therefore the optimum route of being selected by forwarding module may be discontented with the regulation of football association's view.If behind this scheme modifying, select the function of route to bear by routing management, what forwarding was received must be according to the optimum route after certain specific responsibility selection, and forwarding module also need not be preserved the backup of a plurality of routes.
(2) original Max route order can only limit local number of routes and can not limit the far-end route, that be because do not focus on when adding the far-end route local whether exist identical destination route, but directly adding.And whether now because all can import the data structure of depositing adduction relationship of the route of certain VRF and all go here and there on a chained list, being easy to identify local VRF has had this route.If exist, then no longer be subjected to the restriction of max route, if there is no and current number of routes has transfinited, just this node rtbit does not hang on the chain, and later max route becomes big back and only node rtbit need be hung once again and get final product.The processing policy of all these and local route is in full accord.
(3) the derivation route target of the route on the chained list that hangs over first incidence relation (export routetarget) not simultaneously, the route that original realization may can import because of certain VRF is not movable (active) route and do not receive any route.No matter whether whether route be movable (active) state on the chained list of first incidence relation and after adopting this scheme, the capital is suspended to the node rtbit of oneself on the corresponding chain, therefore also with regard to qualified participation import VRF optimum route selection and might finally be chosen as optimum route, thereby even make that certain route is not that optimum route also can import among certain VRF on first incidence relation.
In order to realize said method, aspect following 4, make an amendment:
1) state machine that route is changed is transplanted on the data structure of depositing adduction relationship.Always have 7 kinds of incident: Initialize, Preference, Unreachable, Inactive, Active, Bitset and Bitreset now.The event handling function of their correspondences is rt_event_XXX (XXX is an event name).Each node rtbit can regard the copy of route in specific VRF as now, naturally also can experience the transition of these states, therefore also need events corresponding to handle function, so the event handling function that needs to deposit on the data structure of routing relation is transplanted on the data structure of depositing adduction relationship.The transplanting of these incident functions is revised little, changes the original data structure of depositing routing relation into deposit adduction relationship data structure exactly basically and gets final product.
2) expansion of excellent process is inserted, separated and select to route.Versatile Routing Platform uses in the two-way route chained list of rt_insert function with node rt_entry insert head node rt_head back; With the rt_remove function node rt_entyy is separated from chained list; Use the rt_select_active function from chained list, to select current optimum route sending-down forwarding module.Because, also expanded a doubly linked list of depositing the data structure of adduction relationship among the head node rt_head now, and the chained list of its function and original first incidence relation is identical, therefore, needs above-mentioned three functions also are transplanted on the data structure of depositing adduction relationship.Its inner realization is almost completely the same with original code, and the chained list that just will travel through the data structure of depositing routing relation originally travels through the data structure chained list of depositing adduction relationship instead.Below for the insertion of the node rtbit that increases, separate and select excellent function.
3) modification of route changing process.In changing route, may need some information in the route are preserved so that refreshing the usefulness of route.After will depositing the data structure multiple example type expansion of routing relation now, modification to the data structure of depositing routing relation may not only need original information is kept in the head node rt_head structure of own VRF, and in may needing to be saved in the head node rt_head structure of other VRF.
4) expansion of the data structure of adduction relationship is deposited in distribution.In original realization, only be distribute one deposit the data structure of adduction relationship and with its string on the chained list of first incidence relation.Yet just different now, because the also necessary simultaneously string of node rtbit is on the doubly linked list of the 3rd incidence relation of VRF.Therefore, must arrive the head node rt_head structure whether location among the VRF at place exists identical order IP this moment.If there is no just need call rt_table_add function creation head node rt_head.But in the prior art, need the max route that detects on the corresponding VRF whether to transfinite.If transfinite, then head node rt_head can not create, so node rtbit also can't create, this route just can not import among this VRF.Can realize the restriction of max route by such method to the far-end route naturally.
Though present embodiment forms the 3rd incidence relation by doubly linked list with the existing data structure of depositing the route adduction relationship, is not limited in this, also available other data structures are as long as can associate all routes that allow to import a VRF.As long as realized this primary demand, the routing issue in a plurality of routing tables of being distributed under the MPLS/VPN environment just is converted into and only is present in a routing issue in the routing table under the public network environment so, also just can utilize all flow processs of under the original public network environment route being implemented.
Claims (9)
1, the dual-homed implementation method of a kind of client boundary equipment, described client boundary equipment links to each other with two server's edge devices respectively, on server's edge device, have a plurality of Virtual Switch Instance and with each Virtual Switch Instance corresponding virtual routing table, and the virtual routing table of at least one far-end, routing iinformation to identical destination in same virtual routing table forms first incidence relation, and each routing iinformation can import the formation of all references relation and corresponding second incidence relation of this route of other Virtual Switch Instance; It is characterized in that described method comprises step:
Can import same Virtual Switch Instance and, set up the 3rd incidence relation by described adduction relationship to all routes of identical destination; And
Cause that adding routing iinformation and deletion routing iinformation route selects when excellent again, between the different virtual routing table, carry out overall Route Selection according to described first, second incidence relation and the 3rd incidence relation.
2, the method for claim 1 is characterized in that, described first, second is respectively a two-way chained list with the 3rd incidence relation.
3, method as claimed in claim 1 or 2 is characterized in that described adding routing iinformation comprises step:
After A, native client person edge device are received VPN route from the client person border of far-end, seek the virtual routing table of corresponding far-end in this locality according to the route-distinguisher under the route in the protocol massages, if find then carry out step B, otherwise with this route-distinguisher is that key value (KEY) is created the virtual routing table of new far-end, and continues step B;
B, newly distribute a data structure of depositing route, and fill with the routing iinformation of newly receiving;
C, this data structure of depositing route is inserted in first incidence relation of described far-end routing table, selected excellent rule in this virtual routing table, to carry out route according to route and select excellent;
D, according to derivation route target (export route-tagret) attribute of described route, importing route target (import route-tagte) attribute that mates local all Virtual Switch Instance successively, if can mate, then distribute one to deposit the data structure of route adduction relationship and be inserted in second incidence relation of this route of far-end routing table for Virtual Switch Instance;
E, the head node of in the Virtual Switch Instance corresponding virtual routing table that can import, searching first incidence relation identical with the purpose IP address and the mask of newly-increased route with it;
If do not find, then in current Virtual Switch Instance corresponding virtual routing table, create a new head node, the data structure of depositing the route adduction relationship of distributing among the step D is inserted in corresponding the 3rd incidence relation, selected excellent rule between this virtual routing table, to carry out route according to route and select excellent;
If find, then directly the data structure of depositing the route adduction relationship of distributing among the step D is inserted in corresponding the 3rd incidence relation, select excellent rule between this virtual routing table, to carry out route according to route and select excellent.
4, method as claimed in claim 3 is characterized in that, optimum route changes if route is selected excellent back again, then the optimum route of newly selecting by the corresponding protocol issue.
5, method as claimed in claim 1 or 2 is characterized in that described deletion routing iinformation comprises step:
Route-distinguisher (route-distinguish) attribute that a, basis wait to delete route finds the virtual routing table of depositing this route, and finds the actual data structure of waiting to delete route of depositing;
B, judge whether wait to delete route is current optimum route, if not then carrying out step c, otherwise it is excellent to trigger selecting again of route in first incidence relation, realizes the backup and the switching of route, continues step c;
C, handle and corresponding second incidence relation of this route successively, and trigger node in the 3rd incidence relation and between different virtual routing tables, select excellently again, realize the backup and the switching of route.
6, method as claimed in claim 5 is characterized in that among the step b, if can select new optimum route, then the optimum route of newly selecting by the corresponding protocol issue is cancelled message otherwise notify corresponding protocol to issue.
7, method as claimed in claim 5 is characterized in that:
When not having Routing Protocol to quote in this virtual routing transmitting example to wait to delete route with the knot removal of depositing the route adduction relationship of correspondence;
When all virtual routing transmitting examples are all no longer quoted the data structure deletion that will deposit this routing iinformation when waiting to delete route.
8, the method for claim 1 is characterized in that, the state machine that route changes is arranged on the node in described second incidence relation.
9, the method for claim 1 is characterized in that, to insertion, the deleting mechanism of node in first, second and the 3rd incidence relation, and carries out route and select excellent mechanism to be arranged on the node in described second incidence relation from these incidence relations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031541372A CN100505654C (en) | 2003-08-12 | 2003-08-12 | Method for realizing dual attachment of client boundary device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031541372A CN100505654C (en) | 2003-08-12 | 2003-08-12 | Method for realizing dual attachment of client boundary device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1581828A true CN1581828A (en) | 2005-02-16 |
| CN100505654C CN100505654C (en) | 2009-06-24 |
Family
ID=34580147
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031541372A Expired - Fee Related CN100505654C (en) | 2003-08-12 | 2003-08-12 | Method for realizing dual attachment of client boundary device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100505654C (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007079643A1 (en) * | 2006-01-09 | 2007-07-19 | Huawei Technologies Co., Ltd. | A method and system for realizing duplication based on session border controller |
| CN100411387C (en) * | 2005-03-25 | 2008-08-13 | 杭州华三通信技术有限公司 | Dualhoming network holding method based on RPR network |
| CN100579054C (en) * | 2007-06-11 | 2010-01-06 | 华为技术有限公司 | Controlling method, system and device for reciprocal imputation session boundary controller |
| CN107547378A (en) * | 2017-09-28 | 2018-01-05 | 新华三技术有限公司 | A kind of VPN route learnings method and apparatus |
| CN109445384A (en) * | 2018-10-16 | 2019-03-08 | 劳博(上海)物流科技有限公司 | A kind of more apparatus control systems |
-
2003
- 2003-08-12 CN CNB031541372A patent/CN100505654C/en not_active Expired - Fee Related
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100411387C (en) * | 2005-03-25 | 2008-08-13 | 杭州华三通信技术有限公司 | Dualhoming network holding method based on RPR network |
| WO2007079643A1 (en) * | 2006-01-09 | 2007-07-19 | Huawei Technologies Co., Ltd. | A method and system for realizing duplication based on session border controller |
| CN100508539C (en) * | 2006-01-09 | 2009-07-01 | 华为技术有限公司 | Implement method and system for double-home of session boundary controller |
| US7894410B2 (en) | 2006-01-09 | 2011-02-22 | Huawei Technologies Co., Ltd. | Method and system for implementing backup based on session border controllers |
| CN100579054C (en) * | 2007-06-11 | 2010-01-06 | 华为技术有限公司 | Controlling method, system and device for reciprocal imputation session boundary controller |
| CN107547378A (en) * | 2017-09-28 | 2018-01-05 | 新华三技术有限公司 | A kind of VPN route learnings method and apparatus |
| CN107547378B (en) * | 2017-09-28 | 2020-04-03 | 新华三技术有限公司 | VPN route learning method and device |
| CN109445384A (en) * | 2018-10-16 | 2019-03-08 | 劳博(上海)物流科技有限公司 | A kind of more apparatus control systems |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100505654C (en) | 2009-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI813742B (en) | Asynchronous object manager in a network routing environment | |
| US8750168B2 (en) | Methods and systems to store and forward multicast traffic | |
| KR101989333B1 (en) | Data forwarding method, device and system in software-defined networking | |
| CN103532741B (en) | A kind of access level network topology management method and system | |
| CN1306752C (en) | Using link state information to discover IP network topology | |
| CN105978741A (en) | Network fault handling method and system | |
| CN102142970A (en) | Method for constructing multicast tree in two-layer Ethernet and network node equipment | |
| CN109218046A (en) | The management method and system and storage medium of network slice | |
| CN104219147A (en) | Implementation method and device of VPN (virtual private network) for edge equipment | |
| CN103634423B (en) | Three-layered interface based MPLS-TP (multi-protocol label switching-transport profile) configuration method and device | |
| CN101572669A (en) | Transmitting method of VPN message as well as allocating and deleting method of the router marks thereof | |
| CN103581022B (en) | MAC address finding and transmitting method and device | |
| CN101005458A (en) | Method, device and system for down sending route | |
| CN102546728A (en) | Peer-to-peer network resource downloading method | |
| US8948178B2 (en) | Network clustering | |
| CN108809847A (en) | Realize the method, apparatus and network system of load balancing | |
| US20060268853A1 (en) | Methods and apparatus for distributing label information | |
| CN101616093B (en) | Method and device for realizing user access multi-homing network and network equipment | |
| Son et al. | Forwarding strategy on SDN-based content centric network for efficient content delivery | |
| CN110336730A (en) | A kind of network system and data transmission method | |
| CN106716870A (en) | Local packet switching at a satellite device | |
| CN1581828A (en) | Method for realizing dual attachment of client boundary device | |
| EP1924033B1 (en) | Method and apparatus for implementing a layer1 virtual private network l1 vpn | |
| CN101977123B (en) | Method, system and device for generating virtual private local area network site ID | |
| US6738376B1 (en) | Method and system for managing multicast traffic |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090624 Termination date: 20150812 |
|
| EXPY | Termination of patent right or utility model |