CN104601485A - Network traffic distribution method and routing method for network traffic distribution - Google Patents

Abstract

The invention discloses a network traffic distribution method and a routing method for network traffic distribution. The network traffic distribution method includes: transmitting all or part of traffic of a source node to a target node through multiple chains, with the traffic of any link to the target node being approximate to set traffic of the link. The set traffic of each link is determined according to maximum link usage and link bandwidth. The network traffic distribution method and the routing method for network traffic distribution have the advantages that distributed link traffic of the network traffic is optimally distributed, the maximum link usage of the network is decreased, and by establishing the two-dimensional routing based forwarding method to achieve optimal network traffic distribution, the problem that non-shortest paths of the existing in-domain OSPF (open shortest path first) routing method is solved and the quality of network services is improved.

Description

The distribution method of network traffics and realize the method for routing that network traffics distribute

Technical field

The present invention relates to field of Internet communication, particularly relate to a kind of distribution method of network traffics and realize the method for routing of network traffics distribution.

Background technology

By Business Stream, being mapped on existing physical topology of task is referred to as traffic engineering.Early stage in the Internet, this mapping just maps based on the routing configuration of product simply, but growing along with internet traffic, network operators face huge flow pressure, in order to the network service that the integration that more can effectively utilize network provides high-quality for client, need the traffic engineering setting up complete different implementation.

The traffic engineering method that routing configuration based on product carries out mapping can cause network congestion, is embodied in, some link flow transmission overloads, and other links are relatively idle.Being that in existing territory, the OSPF route of Web vector graphic only carries out route according to shortest path, cannot make full use of non-shortest path by analyzing the reason of known this problem of generation further, thus causing link to use unbalanced.

In sum, a kind of distribution method that can make full use of the network traffics of non-shortest path is needed badly, to solve the problem.

Summary of the invention

One of technical problem to be solved by this invention needs to provide a kind of distribution method that can make full use of the network traffics of non-shortest path.

In order to solve the problems of the technologies described above, the embodiment of the application provide firstly a kind of distribution method of network traffics, comprise: all or part of flow of source node is sent to destination node by multilink, and make the setting flow of flow close to this link of the arbitrary link flowing to described destination node, wherein, the setting flow of each link is the flow value determined according to the bandwidth of maximum link utilization and described link.

Preferably, above-mentioned multilink comprises by source node to the link formed without circulation next-hop node and source node of destination node.

Preferably, linear programming model is utilized to calculate the setting flow of described link, wherein, using first optimization aim of maximum link utilization as linear programming model target function, using second optimization aim of average link utilization as linear programming model target function.

Preferably, linear programming model is as shown in following expression formula:

$\min (\mathrm{\α}+\mathrm{\δ}{\mathrm{\Σ}}_{j\∈V,k\∈E}\left(\frac{g_{\mathrm{jk}}}{\left|E\right|*c_k}\right))$

$\underset{k\∈\mathrm{OUT}\left(i\right)}{\mathrm{\Σ}}{g}_{\mathrm{jk}}-\underset{k\∈\mathrm{IN}\left(i\right)}{\mathrm{\Σ}}{g}_{\mathrm{jk}}=d_{\mathrm{ij}},\∀i\∈V,j\∈V,i\≠j$

$\underset{k\∈\mathrm{OUT}\left(i\right)-\mathrm{Ava}(i,j)}{{\mathrm{\Σg}}_{\mathrm{jk}}}=0,\∀i\∈V,j\∈V,i\≠j$

$\underset{j\∈V}{\mathrm{\Σ}}{g}_{\mathrm{je}}\≤{\mathrm{\αc}}_e,\∀e\∈E$

In formula, V is the set of node, and E is the set of link, | E| is the number of link; α is maximum link utilization, and δ is constant; d _ijrepresent from node i to the uninterrupted of node j; g _jkfor arriving the uninterrupted on the link k of destination node j; What OUT (i) was node i goes out to link set, and what IN (i) was node i enters to link set; Ava (i, j) be node i arrive destination node j without circulation down hop; K and e is the link in set E, c _kfor the bandwidth of link k, c _efor the bandwidth of link e.

On the other hand, the application embodiment still provides a kind of method for routing realizing network traffics and distribute, and comprising: obtain the destination node information corresponding with destination address according to the first mapping table; According to described destination node information inquiry second mapping table, obtain the address block belonging to source address; Address block belonging to source address obtains next-hop node information, and the flow corresponding with source address is sent to next-hop node.

Preferably, the first mapping table comprises destination address prefix and destination node information, and wherein, described destination node information is the address of the memory space corresponding with described destination node pointing to described second mapping table.

Preferably, the second mapping table comprises and divides to source address space the multiple addresses block obtained, and wherein, described address block divides according to the multilink of all or part of flow for forwarding source node.

Preferably, according to following steps, division is carried out to source address space and obtain described multiple addresses block: step one, the link chosen in multilink, wherein, the corresponding address block of the every bar link in multilink; Step 2, flow corresponding for each source address not adding any address block to be added up one by one; Step 3, when cumulative flow and the setting flow close to the link chosen, the source address that executed flow adds up is added the address block corresponding with this link, and again choose a link and return step 2.

Preferably, the first mapping table is stored in TCAM, and the second mapping table is stored in SRAM.

Preferably, the address block belonging to binary search source address is utilized.

Compared with prior art, the one or more embodiments in such scheme can have the following advantages or beneficial effect by tool:

By to utilizing the traffic engineering model shunted without circulation down hop to be optimized, achieving the optimum allocation of the shunting link flow to network traffics, reducing the maximum link utilization of network.Pass through the retransmission method of foundation based on two-dimentional route to realize optimal network assignment of traffic simultaneously, overcome the problem that cannot make full use of non-shortest path in existing territory in OSPF method for routing, improve the quality of network service.

Other advantages of the present invention, target, to set forth in the following description to a certain extent with feature, and to a certain extent, based on will be apparent to those skilled in the art to investigating hereafter, or can be instructed from the practice of the present invention.Target of the present invention and other advantages can by specifications below, claims, and in accompanying drawing, specifically noted structure realizes and obtains.

Although the present invention will be described in conjunction with some exemplary enforcements and using method hereinafter, it will be appreciated by those skilled in the art that, for not being intended to, the present invention will be limited to these embodiments.Otherwise, be intended to cover all substitutes be included in spirit of the present invention and scope that appending claims defines, correction and equivalent.

Accompanying drawing explanation

Accompanying drawing is used to provide the further understanding of technical scheme to the application or prior art, and forms a part for specification.Wherein, the expression accompanying drawing of the embodiment of the present application and the embodiment one of the application are used from the technical scheme explaining the application, but do not form the restriction to technical scheme.

Fig. 1 is the schematic flow sheet realizing the method for routing that network traffics are distributed of the embodiment of the present application;

Fig. 2 is the first mapping table schematic diagram of the embodiment of the present application;

Fig. 3 is the second mapping table schematic diagram of the embodiment of the present application;

Fig. 4 is the core network topological structure schematic diagram of an example;

Fig. 5 (a) is the flow information schematic diagram of an example, and Fig. 5 (b) is the traffic matrix schematic diagram of an example;

Fig. 6 is the first/the second mapping table schematic diagram on the router-A of an example when adopting the method for the embodiment of the present application to forward;

Fig. 7 is the first/the second mapping table schematic diagram on the router D of an example when adopting the method for the embodiment of the present application to forward.

Embodiment

Describe embodiments of the present invention in detail below with reference to drawings and Examples, to the present invention, how application technology means solve technical problem whereby, and the implementation procedure reaching relevant art effect can fully understand and implement according to this.Each feature in the embodiment of the present application and embodiment, can be combined with each other under prerequisite of not conflicting mutually, the technical scheme formed is all within protection scope of the present invention.

Without circulation down hop (Loop-free next hop) also referred to as utilizing down hop, its definition is as shown in expression formula (1):

Dist(N,D)<Dist(N,S)+Dist(S,D) (1)

In formula, the length of the shortest path that what Dist (A, B) represented is from node A to Node B, S represents source node, and D represents destination node, and N represents intermediate node.

The implication of expression formula (1) is, if the length of shortest path that intermediate node N arrives destination node D be less than from source node S arrive the shortest path of destination node D length and arrive from intermediate node N source node S shortest path length and, so just can ensure on intermediate node N, can not be sent to intermediate node N's from source node S, and the flow that finally will arrive destination node D is forwarded back to source node S again, at this moment, can using intermediate node N as from source node S to the route of destination node D without circulation down hop.

Based on the distribution method establishing a kind of network traffics without circulation down hop in the embodiment of the application, be specially, the all or part of flow of source node is sent to destination node (being the Egress node flowing out core net) by multilink, to realize the shunting arriving same destination node, thus reach the object reducing maximum link utilization.Wherein said multilink comprises by source node to the link formed without circulation next-hop node and source node of destination node.

Further, on the basis of known network topological relation and traffic matrix, the distribution of linear programming model to network traffics is utilized to be optimized.Be specially,

Network topology G=<V, E, C, W>, wherein V={1,2 ..., | V|} is the set of node; E={1,2 ..., | E|} is the set of link; C={c _ethe set of link bandwidth, wherein c _erepresent the bandwidth of link e ∈ E; W={w _ethe set of link weight, wherein w _erepresent the weight of link e ∈ E.

Traffic matrix { d _ij, wherein d _ijrepresent from node i to the uninterrupted of node j.Linear programming model is as shown in expression formula (2):

$\min (\mathrm{\&alpha;}+\mathrm{\&delta;}{\mathrm{\&Sigma;}}_{j\&Element;V,k\&Element;E}\left(\frac{g_{\mathrm{jk}}}{\left|E\right|*c_k}\right))---(2-1)$

$\underset{k\&Element;\mathrm{OUT}\left(i\right)}{\mathrm{\&Sigma;}}{g}_{\mathrm{jk}}-\underset{k\&Element;\mathrm{IN}\left(i\right)}{\mathrm{\&Sigma;}}{g}_{\mathrm{jk}}=d_{\mathrm{ij}},\&ForAll;i\&Element;V,j\&Element;V,i\&NotEqual;j---(2-2)$

$\underset{k\&Element;\mathrm{OUT}\left(i\right)-\mathrm{Ava}(i,j)}{{\mathrm{\&Sigma;g}}_{\mathrm{jk}}}=0,\&ForAll;i\&Element;V,j\&Element;V,i\&NotEqual;j---(2-3)$

$\underset{j\&Element;V}{\mathrm{\&Sigma;}}{g}_{\mathrm{je}}\&le;{\mathrm{\&alpha;c}}_e,\&ForAll;e\&Element;E---(2-4)$

In formula (2), V is the set of node, and E is the set of link, | E| is the number of link

The target function that formula (2-1) is linear programming model.In formula, α is maximum link utilization, g _jkfor arriving the uninterrupted on the link k of destination node j, wherein j ∈ V, k ∈ E.Min represents using target function and obtains the optimization aim of minimum value as linear programming model.Further, δ is an enough little constant, so target function deteriorates to min α within the specific limits, by first optimization aim of maximum link utilization as linear programming model; g _jklink flow to be allocated, c _kfor the bandwidth of link k, what represent is average link utilization, after α obtains minimum value, right further be optimized, by second optimization aim of average link utilization as linear programming model.

Formula (2-2), (2-3) and (2-4) are the constraints of linear programming model, wherein,

In formula (2-2), node i that what OUT (i) represented is go out to link set, entering to link set of node i that what IN (i) represented is.The implication of this formula is, flows to the flow of destination node j from source node i, can arrive destination node j by some links.Replace with the shunting mode of multilink and carry out with shortest path the mode that forwards in OSPF/IS-IS network, while alleviating the pressure of shortest path, take full advantage of non-shortest path, make the use of link more balanced.

In formula (2-3), node i that what Ava (i, j) represented is arrive destination node j without circulation down hop.If node i is non-upgrade node (namely adopting shortest path to carry out the node of route in existing OSPF/IS-IS network), the utilized down hop of node i is only the shortest path down hop that node i arrives node j.If node i is upgrade node (namely adopting the method for the embodiment of the present application to carry out the node of route in existing OSPF/IS-IS network), the utilized down hop of node i is that node i arrives the shortest path down hop of node j and other can not cause the down hop of circulation.

It should be noted that, the method of the embodiment of the present application supports partial deployment, namely for the arbitrary node in an OSPF/IS-IS core net, can a part be deployed as adopt existing shortest path carry out route, remainder be deployed as adopt the embodiment of the present application carry out route without circulation down hop.Further, at the flow not corresponding to homologous address of same Nodes, can a part adopt existing shortest path carry out route, what remainder adopted the embodiment of the present application carries out route without circulation down hop.

Formula (2-3) has carried out further restriction to formula (2-2), its implication is, flow to the flow of destination node j from source node i, can only by without circulation down hop formed link, not by can not utilize down hop formation link.It should be noted that shortest path down hop is also without circulation down hop.So, should be 0 at the flow gone out on other link removing the link formed without circulation down hop in link of source node i.Owing to utilizing the link formed without circulation next-hop node and source node to E-Packet, so the circulation of route can not be caused when forwarding.

In formula (2-4), α c _efor the maximum stream flow of link.The implication of this formula is, the flow flowing to arbitrary link of destination node j is not more than the maximum stream flow of link, ensures that the phenomenon of transshipping can not occur link further.

According to the linear programming model of expression formula (2), optimal solution can be tried to achieve in the event expressed by multinomial, and optimal flow assignment does not have route circulation.Namely try to achieve the minimum value of α according to the first optimization aim, try to achieve g according to the second optimization aim _jkminimum value.Wherein, the minimum value of what the minimum value of α represented is maximum link utilization, is the optimization to links all in network, makes the load of all links all be in below the maximum stream flow of link.G _jkminimum value represents is the optimal solution of flow of the link k arriving destination node (i.e. Egress node) j, it can be used as the setting flow of link to distribute the flow in network, make to flow to destination node j the flow of link k close to the setting flow of this link.Be specially, when the flow of the link distributed is greater than or less than the setting flow of this link, the absolute value of the difference of the flow of the link of this distribution and the setting flow of this link all should be enough little, definitely equal surely owing to the flow of the link of distribution and the setting flow of this link can be caused to differ using the flow of an IP address as smallest allocation unit when distributing, such difference allows.

Next, for each destination node j and each link k, need to select corresponding source address, make the source address picked out arrive the flow of destination node j as far as possible close to the setting flow g of link _jk, the embodiment of the application also proposes a kind of method for routing realizing above-mentioned network traffics and distribute for this reason, and as shown in Figure 1, this method for routing comprises:

Step S110, obtain the destination node information corresponding with destination address according to the first mapping table.

Step S120, according to described destination node information inquiry second mapping table, obtain the address block belonging to source address.

Step S130, address block belonging to source address obtain next-hop node information, and the flow corresponding with source address is sent to next-hop node.

Fig. 2 is the first mapping table schematic diagram of the embodiment of the present application.As can be seen from the figure, the first mapping table comprises the corresponding relation of destination address prefix and destination node (i.e. Egress node).The method for building up of the first mapping table is, network topology is regarded as a core net, find out for each object prefix addresses and need from which Egress node of core net to go out to arrive this object prefix addresses, and this corresponding relation is recorded in the first mapping table.Such as object prefix addresses is the destination node of 1.1.1.0/24 is N ₁.It should be noted that, the destination node information in the first mapping table is actually an address pointer, point to certain memory headroom of SRAM, and this memory headroom stores the information relevant with this destination node.Fetched by the address pointer chain realized between the first mapping table and the second mapping table and carry out two-dimentional route.

Fig. 3 is the second mapping table schematic diagram of the embodiment of the present application.As can be seen from the figure, the second mapping table comprises the down hop number corresponding with destination node and the down hop of source address perception.Wherein, the next hop information of source address perception is carried out by source address space dividing the multiple addresses block obtained, and the number of address block is equal with down hop number, if the down hop number of such as destination node has K ₁individual, so according to K ₁source address space is carried out piecemeal, is divided into K ₁block continuous print address space.

According to Fig. 2 and Fig. 3, the concrete routing procedure based on the first mapping table and the second mapping table is:

When router receives a message, in the first mapping table, first search the Egress node (i.e. destination node) corresponding to longest matching prefix of the object IP address of this message.The object IP address of a such as message is 2.2.0.8, and this IP address known belongs to object prefix 2.2.0.0/16, and tabling look-up and obtaining its egress is N1.The N1 obtained is actually the address of a memory headroom of SRAM.Following continuation inquiry second mapping table, deposits the blocking information of the source address space corresponding to this Egress node in the internal storage location that address above mentioned N1 points to.Which block source address space is the source address that further inquiry sends this message be positioned at.Suppose that the source address of this message is positioned at v block (1≤v≤k, wherein k is the number for the multilink E-Packeted) source address space, then the down hop of this message corresponding to v block source address space is forwarded.It should be noted that, searching this process of source address space can be realized by binary chop, and binary chop has simple feature efficiently, and for k bar conversion link, it searches complexity is O (log ₂k).

Further, address block divides according to the multilink of all or part of flow for forwarding source node, and concrete step is:

Step one, the link chosen in multilink, wherein, the corresponding address block of the every bar link in multilink.

Step 2, flow corresponding for each source address not adding any address block to be added up one by one.

Step 3, when cumulative flow and the setting flow close to the link chosen, the source address that executed flow adds up is added the address block corresponding with this link, and again choose a link and return step 2.

Table 1 is the above-mentioned division methods of application carries out piecemeal concrete example to source address space, describes above steps in detail below according to table 1:

Source address space is carried out the method for piecemeal by table 1

As can be known from Table 1, first the embodiment of the present application collects the five-tuple < source address of flow, destination address, source port, destination interface, agreement > information and corresponding uninterrupted information (can see Fig. 5 (a), wherein protocol information is not shown) by NetFlow.Flow Trace information can also be gathered by sFlow, and general router all supports this flow collection function.

Further, according to the topological relation of network, all nodes arriving destination node j are sorted.Be specially, arrive the assignment of traffic { g of destination node j _jk| k ∈ E and g _jklink corresponding to > 0} can form an acyclic oriented topology, sorts, obtain sequence node n to the node under this topology ₁, n ₂..., n _{| V|}.Next, each node n is chosen respectively _u, the source address space of each Nodes is divided.

Node n will be arrived _uand the flow going to destination node j is polymerized according to the IP address of source address space.In this example, the natural order according to IP address sorts, and obtains Z [IP ₁, IP ₂..., IP _n], wherein Z [IP _i] represent No. i-th source address IP _ithe size of corresponding flow.

Definition Two Variables Low and High, is used for representing the scope of source address space.When not carrying out piecemeal, for IPv4 source address space, Low=0.0.0.0, High=255.255.255.255; For IPv6 source address space, Low=0::0, High=ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff.Next, at each node n _uplace to each without circulation next-hop node and this node n _uthe link formed is investigated, and this investigation scope is by k ∈ Out (n _u)-{ k|k ∈ Out (n _u) and g _jk=0} defines.

Deterministic process is, if first judge whether link k is without the last item link in the set of circulation down hop, if so, is then all forwarded by link k by the flow of source address remaining in source address space.If link k be not this without the last item link in the set of circulation down hop, then judge according to the step in above-mentioned method of partition.Be specially, if by node n _uthe source address at place is IP _pflow forwarded by link k, so now by total flow of link k be flow and form acc=acc+Z [IP _p], what wherein acc represented is forward Z [IP _p] before forwarded by link k flow and.Now, if flow and acc are greater than optimal flow assignment g _jk, then continue judgement and add Z [IP _p] whether acc has more afterwards flow is less than does not add Z [IP _p] time less than g _jkflow, if so, then think Z [IP _p] by link k carry out forward the flow of the distribution of this link can be made closer to the setting flow g of link _jk, so be IP by source address _pflow forwarded by link k, and accordingly the initial address of next block source address space is revised as IP _p.Otherwise, if add Z [IP _p] acc has more afterwards flow is more than or equal to and do not add Z [IP _p] time less than g _jkflow, then think not by Z [IP _p] carry out forwarding by link k its flow can be made closer to g _jk, so be IP by source address _pflow by node n _uother links without circulation down hop at place forward, and accordingly the initial address of next block source address space are revised as IP _p-1.

It should be noted that, in other the embodiment of the application, the standard of source address space being carried out to piecemeal can also be acc-g _jk<1/2 [g _jk-(acc-Z [IP _p])] etc. other forms, can choose according to actual conditions, only need meet the flow being sent to each each IP address without circulation next-hop node place and enough little with the absolute value of the difference of the setting flow of this link, definitely equal surely owing to the flow of the link of distribution and the setting flow of this link can be caused to differ using the flow of an IP address as smallest allocation unit when distributing, such difference allows.

Source address space is carried out piecemeal according to without circulation down hop, makes full use of the shunting carrying out flow without circulation down hop link in network topology, overcome the problem that cannot make full use of non-shortest path in existing territory in OSPF method for routing.This method for routing is simple and have effectively achieved all standing of source address space, meets optimum assignment of traffic simultaneously, improves the utilization ratio of link.

Also you need to add is that in addition, based in the network of OSPF/IS-IS, router is transmitted (mapping table) by shortest path first is that each object prefix addresses calculates the shortest path down hop arriving this prefix addresses, and routing table is left in TCAM.The method for routing of the embodiment of the present application carries out two-dimentional route based on two mapping tables, and wherein the first mapping table is stored in TCAM, and the second mapping table is stored in SRAM.Particularly, first revise the first mapping table deposited in TCAM, the port information according to network topology structure and each Nodes sets up corresponding relation.Because the method for routing of the embodiment of the present application supports partial deployment, so only part amendment can be carried out to original route mapping table based on shortest path first.Second mapping table is set up in sram, when the flow in network has greatly changed, is convenient to carry out real-time update to the second mapping table.

The two-dimentional method for routing based on the first/the second mapping table structure of the embodiment of the present application has good autgmentability.When network traffics change, namely traffic matrix changes, only need redistribute link flow according to the method for allocating network flow of the embodiment of the present application, and to set up according to the method for routing of the distribution realizing network traffics and to upgrade the second mapping table be stored in SRAM, without the need to changing the first mapping table, so be more beneficial to expansion with the change adapting to network traffics.

The two-dimentional method for routing based on the first/the second mapping table structure of the embodiment of the present application has the responding ability fast to network traffics change.Simple to operate, the execution speed of TCAM is fast, but the renewal of TCAM is also very complicated simultaneously, and the speed upgrading SRAM is very fast.So in the embodiment of the application, the TCAM being not easy to upgrade is made to keep relative stablizing, when network traffics change, only upgrade the second mapping table deposited in SRAM, and, because the entry number upgraded equals the number of core net node, so it is considerably less to upgrade entry, the change of flow can be adapted to rapidly.

Fig. 4 to Fig. 7 is a concrete example of the method based on the embodiment of the present application, describes in detail below utilize the embodiment of the present application to carry out the process of assignment of traffic and route according to Fig. 4 to Fig. 7.

Fig. 4 is the core network topological structure schematic diagram of an example.As can be seen from the figure, this network has 6 nodes, supposes that the bandwidth of every bar link is all 100, and the weight of every bar link is all 1, then can extract following topological relation according to Fig. 4: node set V={A, B, C, D, E, F}, link set E={ (A, B), (B, C), (C, F), (A, D), (D, E), (D, F), (E, F) }, link bandwidth set C={c _ij=100|i ∈ V, j ∈ V}, link weight set W={w _ij=1|i ∈ V, j ∈ V}.

Fig. 5 (a) is the flow information schematic diagram of an example, and Fig. 5 (b) is the traffic matrix schematic diagram of an example.List in Fig. 5 (a) each message flow information and the source address of correspondence.These flow informations are polymerized according to source node, destination node, traffic matrix figure { d can be obtained _ij, as shown in Fig. 5 (b).As can be seen from Fig. 5 (b), the total flow of node A to node F is 50, and the total flow of node D to node F is 50, and the total flow of node C to node F is 20.Further analysis is known, if according to Shortest path routing, so node A can arrive node F via the path of A → D → F, so the flow on link D → F will be 50+50, namely now the link utilization of link (D, F) will reach 100%.For ensureing the efficient communication of network, the link utilization that this is maximum must be reduced.

Next according to Fig. 4 determine each node of this core network arrive destination node F without circulation down hop.Weight due to every bar link is all 1, so can being respectively without circulation down hop in the hope of each node: Ava (A, F)={ D, B}, Ava (B, F)={ C, A}, Ava (C, F)={ F}, Ava (D, F)={ F, E}, Ava (E, F)={ F, D}, and be easy to obtain B and A each other without circulation down hop, D and E is each other without circulation down hop.By the above-mentioned relation obtained and the topological relation obtained according to Fig. 4, and according to the traffic matrix that Fig. 5 obtains, the linear programming model inputting the embodiment of the present application together carries out assignment of traffic.

By calling being assigned as of the link flow of the optimum under the traffic matrix shown in Fig. 5 (b) that linear programming model obtains: g _{f, A → B}=20; g _{f, B → C}=20; g _{f, C → F}=40; g _{f, A → D}=30; g _{f, D → F}=40; g _{f, D → E}=40; g _{f, E → F}=40.Other links arrive destination node F does not have flow, can find out that maximum link utilization now reduces greatly.

The schematic diagram of the first/the second mapping table on router-A when adopting the method for the embodiment of the present application to forward and on router D is respectively shown in Fig. 6 and Fig. 7.In figure 6, the structure on the left side is the first mapping table, and in the core network of this example, destination address has 9.9.9.1 and 9.9.9.2, and their destination node is F.Structure on the right of Fig. 6 is the second mapping table, according to analysis above, at node A place, what arrive destination node F has two without circulation next-hop node, be respectively node D and Node B, so in the second mapping table shown in Fig. 6, the down hop number corresponding to destination node F is 2, utilizes the method for partition in the embodiment of the present application source address space to be divided into continuous print address, two addresses block.

Further can also according to g _{f, A → B}=20 and g _{f, A → D}=30 pairs of piecemeal results are verified.Known according to Fig. 5 (a) and Fig. 5 (b), at node A place, the flow of source address 1.1.1.1 and 2.2.2.2 is forwarded by node D, then the total flow of link A → D is 30, the flow of source address 3.3.3.3 is forwarded by Node B, then the total flow of link A → B is 20, meets the result of optimal flow assignment.

Conclusion and Fig. 6 of Fig. 7 are similar, repeat no more.Other node arrives destination node F and only has a down hop, also repeats no more herein.

The traffic engineering method based on two-dimentional route that the present invention proposes not only can optimized network assignment of traffic, also there is good autgmentability, when network traffics change, re-routed by the information upgraded fast in SRAM, and be common to IPv4 and IPv6 network and support partial deployment.

Although the execution mode disclosed by the present invention is as above, the execution mode that described content just adopts for the ease of understanding the present invention, and be not used to limit the present invention.Technical staff in any the technical field of the invention; under the prerequisite not departing from the spirit and scope disclosed by the present invention; any amendment and change can be done what implement in form and in details; but scope of patent protection of the present invention, the scope that still must define with appending claims is as the criterion.

Claims (10)

1. a distribution method for network traffics, comprising:

The all or part of flow of source node is sent to destination node by multilink, and makes the setting flow of flow close to this link of the arbitrary link flowing to described destination node, wherein,

The setting flow of each link is the flow value determined according to the bandwidth of maximum link utilization and described link.

2. method for allocating network flow according to claim 1, is characterized in that, described multilink comprises by source node to the link formed without circulation next-hop node and source node of destination node.

3. method for allocating network flow according to claim 1 and 2, it is characterized in that, linear programming model is utilized to calculate the setting flow of described link, wherein, using first optimization aim of maximum link utilization as linear programming model target function, using second optimization aim of average link utilization as linear programming model target function.

4. method for allocating network flow according to claim 3, is characterized in that, described linear programming model is as shown in following expression formula:

$\min (\mathrm{\&alpha;}+\mathrm{\&delta;}{\mathrm{\&Sigma;}}_{j\&Element;V,k\&Element;E}\left(\frac{g_{\mathrm{jk}}}{\left|E\right|*c_k}\right))$

$\underset{k\&Element;\mathrm{OUT}\left(i\right)}{\mathrm{\&Sigma;}}{g}_{\mathrm{jk}}-\underset{k\&Element;\mathrm{IN}\left(i\right)}{\mathrm{\&Sigma;}}{g}_{\mathrm{jk}}=d_{\mathrm{ij}},\&ForAll;i\&Element;V,j\&Element;V,i\&NotEqual;j$

$\underset{k\&Element;\mathrm{OUT}\left(i\right)-\mathrm{Ava}(i,j)}{\mathrm{\&Sigma;}}{g}_{\mathrm{jk}}=0,\&ForAll;i\&Element;V,j\&Element;V,i\&NotEqual;j$

$\underset{j\&Element;V}{\mathrm{\&Sigma;}}{g}_{\mathrm{je}}\&le;\mathrm{\&alpha;}{c}_e,\&ForAll;e\&Element;E$

In formula, V is the set of node, and E is the set of link, | E| is the number of link; α is maximum link utilization, and δ is constant; d _ijrepresent from node i to the uninterrupted of node j; g _jkfor arriving the uninterrupted on the link k of destination node j; What OUT (i) was node i goes out to link set, and what IN (i) was node i enters to link set; Ava (i, j) be node i arrive destination node j without circulation down hop; K and e is the link in set E, c _kfor the bandwidth of link k, c _efor the bandwidth of link e.

5. realize the method for routing that network traffics are distributed, comprising:

The destination node information corresponding with destination address is obtained according to the first mapping table;

According to described destination node information inquiry second mapping table, obtain the address block belonging to source address;

Address block belonging to source address obtains next-hop node information, and the flow corresponding with source address is sent to next-hop node.

6. method for routing according to claim 5, it is characterized in that, described first mapping table comprises destination address prefix and destination node information, and wherein, described destination node information is the address of the memory space corresponding with described destination node pointing to described second mapping table.

7. method for routing according to claim 5, it is characterized in that, described second mapping table comprises and divides to source address space the multiple addresses block obtained, and wherein, described address block divides according to the multilink of all or part of flow for forwarding source node.

8. method for routing according to claim 7, is characterized in that, carries out division obtain described multiple addresses block according to following steps to source address space:

Step one, the link chosen in multilink, wherein, the corresponding address block of the every bar link in multilink;

Step 2, flow corresponding for each source address not adding any address block to be added up one by one;

Step 3, when cumulative flow and the setting flow close to the link chosen, the source address that executed flow adds up is added the address block corresponding with this link, and again choose a link and return step 2.

9. the method for routing according to any one of claim 5 to 7, is characterized in that, described first mapping table is stored in TCAM, and described second mapping table is stored in SRAM.

10. method for routing according to claim 5, is characterized in that, utilizes the address block belonging to binary search source address.