CN102136998A - Traffic engineering and server selection joint optimization method, system and related equipment - Google Patents
Traffic engineering and server selection joint optimization method, system and related equipment Download PDFInfo
- Publication number
- CN102136998A CN102136998A CN2010102701012A CN201010270101A CN102136998A CN 102136998 A CN102136998 A CN 102136998A CN 2010102701012 A CN2010102701012 A CN 2010102701012A CN 201010270101 A CN201010270101 A CN 201010270101A CN 102136998 A CN102136998 A CN 102136998A
- Authority
- CN
- China
- Prior art keywords
- link
- node
- virtual
- traffic
- user 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
- 238000005457 optimization Methods 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000004364 calculation method Methods 0.000 claims abstract description 19
- 230000006870 function Effects 0.000 claims description 36
- 238000000638 solvent extraction Methods 0.000 claims description 14
- 230000011218 segmentation Effects 0.000 claims description 10
- 238000009826 distribution Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 17
- 230000009977 dual effect Effects 0.000 description 15
- 238000012546 transfer Methods 0.000 description 6
- 239000013256 coordination polymer Substances 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 4
- 238000005192 partition Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000004904 shortening Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 241000287196 Asthenes Species 0.000 description 1
- 241000764238 Isis Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- COCAUCFPFHUGAA-MGNBDDOMSA-N n-[3-[(1s,7s)-5-amino-4-thia-6-azabicyclo[5.1.0]oct-5-en-7-yl]-4-fluorophenyl]-5-chloropyridine-2-carboxamide Chemical compound C=1C=C(F)C([C@@]23N=C(SCC[C@@H]2C3)N)=CC=1NC(=O)C1=CC=C(Cl)C=N1 COCAUCFPFHUGAA-MGNBDDOMSA-N 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002040 relaxant effect Effects 0.000 description 1
- 229910052814 silicon oxide Inorganic materials 0.000 description 1
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The embodiment of the invention provides a traffic engineering and server selection joint optimization method, a traffic engineering and server selection joint optimization system and related equipment. An optimal solution is calculated by utilizing a small amount of information in a relatively shorter convergence time on the premise of not disclosing key bottom-layer network information to a content provider (CP) by an instant service provider (ISP). The method comprises that: a router determines each outgoing link bearing traffic belonging to the same user node; the router acquires optimal link weights of each outgoing link; and the router acquires determines a division strategy of the traffic belonging to the same user node on any outgoing link (u,v) in the outgoing links according to the optimal link weights and the traffic belonging to the same user node. In the method, the system and the related equipment, calculation is not required to be performed for a plurality of turns, so the amount of transmitted information is greatly reduced, the convergence time of the system is shortened, time consumption is low generally, and the traffic division determining efficiency is higher.
Description
Technical Field
The invention relates to the field of communication, in particular to a method, a system and related equipment for joint optimization of traffic engineering and server selection.
Background
Internet Service Providers (ISPs) provide Internet connectivity via their own physical networks. Since there are multiple paths between a source node and a destination node in a network topology, when an ISP plans to send a certain Traffic from a certain source node to a certain destination node, it needs to determine how to distribute the Traffic on each path, so as to optimize the overall performance (e.g., delay, load balancing, etc.) of the network.
And Content Providers (CP) provide services such as sharing of files, audio and video, etc. to their users in terms of Content using internet connections. Unlike ISPs, CPs have no authority to change the routing of the underlying network. For each user request, the CP needs to decide how to distribute the traffic requested by the user to different servers, and the above problem that the CP needs to solve is called a Server Selection (SS) problem.
Generally, the ISP and CP solve TE and SS independently, respectively, and there is no information that can be shared between them. Since both TE and SS affect the network conditions, both ISP and CP need to adjust TE and SS continuously until the system reaches a stable state. As mentioned above, since there is generally no information exchange between ISP and CP, under an uncooperative scenario, CP needs to rely on probing end-to-end and other underlying network information to resolve SS; in some collaboration scenarios, the ISP provides accurate underlying network information, such as topology information, link state information, and routing information, to the CP. However, since there is always an inconsistency between the optimization goals of ISP to TE and CP to SS, both sides would like to optimize their goals from their own standpoint. Therefore, in the non-cooperative scenario or the partially cooperative scenario, the non-cooperative gaming mechanism of TE and SS can only ensure that the system converges to a suboptimal balance, and the system performance cannot reach Pareto (Pareto) optimum.
In order to ensure that the system converges to pareto optimality, the ISP and CP need to implement joint optimization between TE and SS. In the joint optimization, the ISP and the CP realize complete information sharing, that is, the ISP provides the CP with its own network topology information, link state information, routing information, etc., and the CP provides the ISP with its own server information, user requirement information, etc. When the two realize complete information sharing, TE and SS are not solved independently. Joint optimization is performed simultaneously for TE and SS, rather than in an iterative fashion (i.e., ISP performs TE once, CP performs SS once more, and so on). The joint optimization can achieve good effect on a plurality of applications, such as on-demand services and the like.
However, one of the challenges faced by the joint optimization of TE and SS is how to design a fast convergence protocol to ensure that the system achieves the best performance while exposing as little information as possible between ISP and CP to each other. For example, the ISP exposes as little network topology information, link state information, routing information, etc. as possible, while the CP exposes as little server information, user demand information, etc. as possible.
In view of the above-mentioned problems faced by joint optimization of TE and SS, the industry proposes a joint SS and TE (COST, COoperative SS and TE) protocol. COST uses dual decomposition to decompose the initial joint optimization problem into two sub-optimization problems similar to TE and SS and one main optimization problem. The two sub-optimization problems are mutually related through a common dual variable, and the main optimization problem is responsible for continuously updating the related dual variable so as to enable the system to approach the optimal solution. In solving the joint optimization problem of TE and SS, given the associated dual variables, the ISP solves the TE-like sub-optimization problem TE-NBS, while the CP solves the SS-like sub-optimization problem SS-NBS. After each round of optimization, the system updates the associated dual variables for the next round of joint optimization by solving the primary optimization problem according to the solutions of the two sub-optimization problems. After a sufficient number of rounds of optimization, COST ensures that optimal system performance can be achieved.
In a specific implementation, the ISP can utilize existing techniques to solve TE-like sub-optimization problem TE-NBS. Techniques that may be employed include multiprotocol Label Switching (MPLS) with centralized algorithms and exponentially penalized traffic Partitioning (PEFT) with distributed algorithms, since it is required to obtain the optimal solution to the sub-optimization problem. Meanwhile, the CP needs to get accurate underlying network information (including topology and time delay) to solve the SS-like sub-problem SS-NBS.
Since COST needs to be optimized through multiple rounds, ISP needs to solve the sub-optimization problem TE-NBS similar to TE in each round of optimization, CP needs to solve the sub-optimization problem SS-NBS similar to SS, and communication is realized between the two by updating the price of each link in the network, compared with the existing TE protocol, the time consumption of COST is multiplied (proportional to the number of rounds of optimization), and further, the information transfer required by COST not only includes the information transfer for solving the two sub-optimization problems in each round of optimization, but also includes the information transfer required for updating the associated dual variables when solving the main optimization problem. That is, COST requires a long convergence time and a large amount of information transfer in reaching the optimal solution.
Furthermore, COST requires the ISP to provide critical underlying network information (e.g., network topology information, link state information, etc.) to the CP when solving the SS optimization problem. But many times such a requirement is not reasonable for an ISP, especially when the ISP dominates the collaboration; when multiple CPs are included in the network, the ISP is also willing to leak critical underlying network information to each CP.
Disclosure of Invention
The embodiment of the invention provides a combined optimization method, a system and related equipment for traffic engineering and server selection, which can obtain an optimal solution by short convergence time and a small amount of information transmission on the premise of not leaking key underlying network information to a CP (provider service provider).
The embodiment of the invention provides a combined optimization method for traffic engineering and server selection, which comprises the following steps: the router determines each outward link bearing the flow belonging to the same user node;
the router acquires optimal link weights of the outward links, wherein the optimal link weights are obtained by solving a network entropy maximization NEMR problem of the network topology with the added virtual nodes;
and the router determines a segmentation strategy of the traffic belonging to the same user node on any one of the outward links (u, v) according to the optimal link weight and the traffic belonging to the same user node, wherein u represents a node where the router is located, and v represents another node forming the outward link (u, v).
The embodiment of the invention provides a combined optimization method for traffic engineering and server selection, which comprises the following steps: the method comprises the steps that a server acquires the shortest path length from a router connected with the server to a user node and the equivalent number from the router to the user node;
the server acquires optimal link weights of all virtual links, wherein the optimal link weights are obtained by solving a network entropy maximization NEMR problem of a network topology with added virtual nodes, and the virtual links are formed by the server with the added virtual nodes in the network topology;
and the server determines the flow demand distributed to the user node on the server according to the optimal link weight, the shortest path length, the equivalent number and the total flow demand belonging to the user node.
The embodiment of the invention provides a combined optimization method for traffic engineering and server selection, which comprises the following steps: the method comprises the steps that a virtual node is added on a network topology, the flow demand from each server node to a user node on the network topology is converted into the virtual flow from the virtual node to the user node, the virtual flow is constantly equal to the total flow demand of the user node, and a virtual link is formed between the virtual node and each server node on the network topology;
the virtual flow is taken as a constant of a user demand constraint item in a multi-item network flow MCF problem, the MCF problem is solved to obtain an optimal flow on each link of the network topology, each link of the network topology comprises a physical link and the virtual link on the network topology, and the user demand constraint item in the MCF problem is as follows:
wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set, respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), and D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand;
solving a network entropy maximization NEMR problem according to the optimal flow on the link to obtain optimal link weights on the virtual link and the physical link;
distributing the optimal link weights to router nodes on the network topology.
An embodiment of the present invention provides a router, including: an outward link determining module, configured to determine outward links carrying traffic belonging to the same user node;
an obtaining module, configured to obtain an optimal link weight of each outbound link determined by the outbound link determining module, where the optimal link weight is obtained by solving a network entropy maximization NEMR problem in which a network topology of a virtual node is increased;
and a partitioning policy determining module, configured to determine, according to the optimal link weight and the traffic belonging to the same user node, a partitioning policy of the traffic belonging to the same user node on any one of the outbound links (u, v), where u denotes a node where the router is located, and v denotes another node constituting the outbound link (u, v).
An embodiment of the present invention provides a server, including: the first acquisition module is used for acquiring the shortest path length from the router to a user node and the equivalent number from the router to the user node from the router connected with the server;
a second obtaining module, configured to obtain an optimal link weight of each virtual link, where the optimal link weight is obtained by solving a problem of network entropy maximization NEMR of a network topology to which a virtual node Ns is added, and the virtual link is formed by the virtual node added in the network topology and the server;
and the flow determining module is used for determining the flow demand distributed to the user nodes on the server according to the total flow demand belonging to the user nodes, the optimal link weight obtained by the second obtaining module, the shortest path length obtained by the first obtaining module and the equivalent number from the router to the user nodes.
The embodiment of the invention provides an optimal link weight obtaining device, which comprises: the conversion module is used for converting the flow demand from each server node to a user node on the network topology into the virtual flow from the virtual node to the user node by adding the virtual node on the network topology, wherein the virtual flow is constantly equal to the total flow demand of the user node, and a virtual link is formed between the virtual node and each server node on the network topology;
an optimal flow solving module, configured to use the virtual flow as a constant of a user demand constraint term in a multi-commodity network flow MCF problem, and solve the MCF problem to obtain an optimal flow of each link of the network topology, where each link of the network topology includes a physical link and a virtual link on the network topology, and the user demand constraint term in the MCF problem is:
wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set on the network topology G (V, E), respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), and D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand;
the optimal link weight solving module is used for solving a network entropy maximization NEMR problem according to the optimal flow solved by the optimal flow solving module on the link, so as to obtain optimal link weights on the virtual link and the physical link;
a distribution module for distributing the optimal link weight to a router node on the network topology.
The embodiment of the invention provides a combined optimization system for traffic engineering and server selection, which comprises the following steps: the system comprises an optimal link weight acquisition device, a server and a router;
the optimal link weight obtaining device is used for adding virtual nodes on a network topology to enable each server node on the network topology to reach a userThe flow demand of a node is converted into virtual flow from the virtual node to the user node, the virtual flow is always equal to the total flow demand of the user node, a virtual link is formed between the virtual node and each server node on the network topology, the virtual flow is taken as a constant of a user demand constraint item in a multi-commodity network flow MCF problem, the MCF problem is solved to obtain the optimal flow of each link of the network topology, each link of the network topology comprises a physical link and a virtual link on the network topology, and the user demand constraint item in the MCF problem is as follows:wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set, respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand, solving a network entropy maximization NEMR problem according to the optimal traffic on the link to obtain optimal link weights on the virtual link and the physical link, and distributing the optimal link weights to router nodes on the network topology;
the server is used for acquiring the shortest path length from the router to a user node and the equivalent number from the router to the user node from the router connected with the server, acquiring the optimal link weight of each virtual link, wherein the optimal link weight is obtained by solving a network entropy maximization NEMR problem of a network topology with the added virtual nodes, the virtual links are formed by the added virtual nodes in the network topology and the server, and the traffic demand distributed to the user node on the server is determined according to the optimal link weight, the shortest path length, the equivalent number from the router to the user node and the total traffic demand belonging to the user node;
the router is configured to determine each outbound link carrying traffic belonging to the same user node, obtain an optimal link weight of each outbound link, where the optimal link weight is obtained by solving a network entropy maximization problem NEMR that increases a network topology of a virtual node, and determine, according to the optimal link weight and the traffic belonging to the same user node, a partitioning policy of the traffic belonging to the same user node on any one outbound link (u, v) in the outbound links, where u denotes a node where the router is located, and v denotes another node constituting the outbound link (u, v).
As can be seen from the foregoing embodiments of the present invention, a router can determine a partitioning policy of traffic belonging to the same user node on any one of the outbound links according to the optimal link weight and the traffic belonging to the same user node by obtaining the optimal link weight of each outbound link. Therefore, the router in the above embodiment is equivalent to only solving a sub-optimization problem similar to TE, and compared with the prior art, the router does not need multiple rounds of calculation, thereby greatly reducing the amount of information transfer, shortening the convergence time of the system, reducing the overall time consumption, and having higher efficiency of determining traffic segmentation.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the prior art or the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive labor.
FIG. 1 is a flow chart of a method for joint optimization of traffic engineering and server selection according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a directed link in a topological network diagram provided in an embodiment of the present invention;
fig. 3 is a schematic diagram of a path from a node u to a user node t according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating a method for joint optimization of traffic engineering and server selection according to another embodiment of the present invention;
fig. 5 is a schematic diagram of a network topology after adding a virtual node according to an embodiment of the present invention;
FIG. 6 is a flowchart illustrating a method for joint optimization of traffic engineering and server selection according to another embodiment of the present invention;
fig. 7 is a schematic diagram of a logical structure of a router according to an embodiment of the present invention;
fig. 8 is a schematic diagram of a logical structure of a router according to another embodiment of the present invention;
FIG. 9 is a schematic diagram of a logical structure of a server according to an embodiment of the present invention;
FIG. 10 is a schematic diagram of a logical structure of a server according to another embodiment of the present invention;
fig. 11 is a schematic diagram of a logical structure of an optimal link weight obtaining apparatus according to an embodiment of the present invention;
fig. 12 is a schematic logical structure diagram of a joint optimization system for traffic engineering and server selection according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, a flow diagram of a joint optimization method for traffic engineering and server selection according to an embodiment of the present invention is shown, which mainly includes the following steps:
s101, the router determines each outward link bearing the flow belonging to the same user node.
So-called outbound Link, in a directed graph G (V, E) (e.g., a network topology graph, etc., where V represents a set of nodes in the directed graph and E represents a set of links in the directed graph), there may be such a Link: for any node, if traffic is flowing from the node and into a neighbor node of the node, the link between the node and such neighbor node is the node's outbound link. As shown in fig. 2, if the direction of the arrow indicates the flow of data (traffic), node u and neighboring node V1Neighbor node V2.nFormed directed links (V)1U) directed link (u, V)2) And directed links (u, V)n) Since the data flows from node u, from neighbor node V2And a neighboring node VnIncoming, then there is a directed link (u, V)2) And a directed link (u, V)n) That is, the outward link of node u, and the directional link (V)1And u) is not an outbound link for node u.
In a network topology, for a given router (a node in the network topology), there are multiple links formed by the router. Therefore, it is necessary to determine which are the outgoing links carrying traffic belonging to the same user node. In the embodiment of the present invention, the router may determine each outbound link carrying traffic belonging to the same user node by looking up network topology information such as a routing table.
S102, the router obtains the optimal link weight of each outward link.
In the embodiment of the present invention, the optimal link weight is obtained by solving a Network Entropy Maximization problem (NEMR) that increases the Network topology of the virtual node. The solution of the NEMR problem can be seen in the embodiment shown in fig. 4 below. The solution of the NEMR problem is the optimal link weights for each link (including the outbound links) in the network topology map, which can be distributed to each router, which then obtains the optimal link weights for each outbound link.
The optimal link weight of the outbound link may have other solving and distributing manners, and is not limited to the embodiment shown in fig. 4.
S103, determining a segmentation strategy of the traffic belonging to the same user node on any one outward link (u, v) in each outward link according to the optimal link weight and the traffic belonging to the same user node.
The user node may be a final destination node of traffic, for example, a terminal such as a Personal Computer (PC) or a mobile phone in an actual network; or a general destination node, which aggregates traffic allocated to several users of the lower layer.
Hereinafter, it will be described how a router determines a partitioning policy of traffic belonging to the same user node t on any one of the outbound links (u, v) by using a network topology diagram in which a node u represents an actual router, a node t represents a user node, and an arrow direction represents an inbound and outbound node flow direction of data (traffic).
As shown in fig. 3, it is a schematic diagram of a path from a node u to a user node t. Assume that any of these paths through node v to user node t is path u-v-c- · -t in the graph. As explained above, link (u, v) is an outbound link to node u, node bPoint v is another node of the link (u, v). The router can easily obtain the shortest path from the router to the user node t through the outward link (u, v), and the shortest path is assumed to be u-v-c-d-t; further, the router can also obtain the length of the shortest path u-v-c-d-t, and the length of the shortest path from the router to the user node t through any one outward link (u, v) is recorded asHere, w (u, v) represents a link weight (also a length of the link) of the link (u, v), and this link weight is also an optimal solution of the NEMR problem described above, i.e., an optimal link weight. Similarly, the router can easily obtain its shortest path to the user node t (it should be understood by those skilled in the art that the shortest path may also be the shortest path from the router to the user node t via the outbound link (u, v), for example, the path u-v-c-d-t in this embodiment, assuming that this shortest path is u-e-f-t; further, the router can also obtain the length of the shortest path u-e-f-t, and the length of the shortest path from the router to the user node t is recorded asThen, calculateAnddifference of (2), the difference isIs shown, i.e.
Then, based on the differenceAnd the equivalent number from the router to the user node t, and determining any one of the flows belonging to the same user node t in each outward linkSplit ratio on bar outbound links (u, v). An example of calculating this division ratio is given below.
S1, calculating the product of the router equivalent number and a decay parameter, wherein the decay parameter is the difference of the natural exponent eInverse of the power, i.e. calculationObtaining a traffic splitting function for any of said outbound links (u, v)Is shown, i.e.Wherein,is an equivalent number from the router to the user node t, the value of which is defined asIn the expression, the expression is given,is the ith path length from node v to user node t,is the shortest path length from node v to user node t. Flow division functionCorresponding to any of the above mentioned outbound links (u, v).
It should be noted that the router may send the shortest path length to the user node t and the equivalent number to the user node t to the server connected to the router at the request of the server.
S2, a sum of the traffic splitting functions corresponding to the respective outbound links is calculated.
Similar to the calculation of the traffic splitting function of the outbound link (u, v) in S1, the traffic splitting function corresponding to each outbound link may be calculated using the same method, i.e., calculatingWherein, (u, j) ∈ E represents any one of the outbound links (u, j) in the outbound link set of the node u, and then summing, i.e. calculating
After the division ratio of the traffic belonging to the same user node t on any one of the outward links (u, v) of the nodes u is obtained through the calculation of S3, the product of the traffic belonging to the same user node t and the division ratio is calculated, that is, the product of the traffic belonging to the same user node t and the division ratio is calculatedIn the expression, the expression is given,representing traffic flowing from node u to the same user node t. Traffic belonging to the same user node is used for split traffic on any of the above-mentioned outgoing links (u, v)Is shown, i.e.
In the embodiment shown in fig. 1, the router may determine the partitioning policy of the traffic belonging to the same user node on any one of the outbound links according to the optimal link weight and the traffic belonging to the same user node by obtaining the optimal link weight of each outbound link. Therefore, the router in the above embodiment is equivalent to only solving a sub-optimization problem similar to TE, and compared with the prior art, the router does not need multiple rounds of calculation, thereby greatly reducing the amount of information transfer, shortening the convergence time of the system, reducing the overall time consumption, and having higher efficiency of determining traffic segmentation.
In a practical network, since the traffic forwarded from one node a is forwarded through several intermediate nodes, the traffic may return to the node a again, that is, the traffic may encounter a "routing loop" when being routed in the network. In order to avoid the occurrence of routing loops, in the embodiment of the present invention, when selecting the outward links, the outward links are preferably selected such that the shortest path length from each node in the respective outward links to the user node t is smaller than the shortest path length from the router (node u) to the user node t. That is, at each router (node u), traffic is always split across those nodes closer to the destination node. Therefore, in the embodiment of the present invention, the traffic splitting function on any one of the outbound links (u, v) is expressed as:
referring to fig. 4, a flow chart of a joint optimization method for traffic engineering and server selection according to another embodiment of the present invention is shown, which mainly includes the following steps:
s401, by adding virtual nodes on the network topology, the flow demand from each server node to the user node on the network topology is converted into the virtual flow from the virtual node to the user node.
In this embodiment, the added virtual nodes may form virtual links with the server nodes on the network topology. As shown in FIG. 5, after adding the virtual node Ns, each server node (S)1、S2、..、Sn) The sum of Traffic demands (Traffic Demand) to the user node t is equal to the virtual Traffic from the virtual node Ns directly to the user node t, and is represented by D (Ns, t). The virtual traffic is assumed to be a traffic that is not actually present from the virtual node Ns to the user node t, and is added to the virtual node Ns to solve the problem. Although the traffic demand on each (Si, t) pair (where i is 1, 2,.. times.n) varies for the customer node t, the sum of the traffic demands on the respective (Si, t) pairs is always equal to the total traffic demand of the customer node t, and therefore the virtual traffic of the virtual node Ns directly to the customer node t is a constant which is always equal to the total traffic demand of the customer node t.
S402, taking the virtual Flow as a constant of a user demand constraint item in a Multi-commodity network Flow (MCF) problem, and solving the MCF problem to obtain the optimal Flow of each link of the network topology.
The MCF problem is a problem expressed by the following expression:
constraint conditions are as follows:
variables are as follows: <math><mrow><msubsup><mi>f</mi><mrow><mi>u</mi><mo>,</mo><mi>v</mi></mrow><mi>t</mi></msubsup><mo>≥</mo><mn>0</mn></mrow></math>
in the MCF problem, [ phi ] (f)u,v,cu,v) Represents the cost of the link (u, v), which is the link traffic (also called link load) fu,vAnd link capacity cu,vAs a function of (c). The link cost Φ is usually chosen as a convex function, e.g. the cost of a link is taken as the utilization f of the linku,v/cu,vOr a piecewise-linear function of link utilization, the objective of solving the MCF problem is to minimize the sum of the costs of all links. Constraint term 1 indicates the flow f of the link (u, v)u,vEqual to the traffic of all the different users passing through the linkSumming; constraint entry 2 indicates that the traffic of the link (u, v) cannot exceed the capacity c of the linku,v(ii) a Constraint term 3 represents a user demand constraint that satisfies the traffic conservation requirement, where D (s, t) represents the traffic demand from server s to user t. In the TE problem, D (s, t) is a constant. The MCF problem is a typical convex optimization problem, which can obtain an optimal solution within polynomial time, i.e. the optimal flow f of each linku,v。
Conventional TE considers how, at each node, a router distributes the passing traffic onto outbound links to flow to neighboring nodes. It is worth noting that conventional TE assumes that the traffic demand between the source node and the destination node is constant, i.e. the traffic matrix (consisting of individual D (s, t)) is constant.
Before the network topology is not added with virtual nodes, as shown in fig. 6, each node Si (e.g., server node) transmits a traffic demand D(s) to the user node tiAnd t) includes: indicating CP traffic D transmitted from node Si to user node tcp(siT) and background traffic D delivered to the user tbg(siT), background flow Dbg(siT) is a constant, Dcp(siAnd t) is a variable in the SS problem.
However, for joint optimization of TE and SS, traffic demand D(s) of user node tiT) is no longer a constant because of the CP traffic D per node Si to the subscriber tcp(siT) is a variable in the SS problem. Despite the CP traffic D per node Si to the subscriber tcp(siT) is a variable in the SS problem, however, for a certain user node t, the total traffic demand is a constant, i.e. the sum of the CP traffic provided by all nodes Si serving the user node tIs a constant.
After adding the virtual node Ns and the virtual link in the network topological graph, the CP flow D of a single server to a single user changescp(siT) into a total CP traffic for the user that is constant for the user for the plurality of servers serving the userDue to Dbg(siT) is a constant, and after the virtual node Ns and the virtual link are added in the network topological graph, the single server has a variable traffic demand D(s) for a single useriT) is converted into a plurality of servers serving the user for the userFor a constant total flow demand D (s, t).
As shown in step S401, after the virtual node Ns is added to the network topology, the virtual traffic D (Ns, t) from the virtual node Ns to the user node t is a constant, and the virtual traffic D (Ns, t) is equal to the total traffic demand D (S, t) of the user node t. That is, after adding the virtual node Ns, the traffic demand D(s) of the individual user due to the individual serveriT) is changed so that the constraint of the MCF problem is not satisfied, i.e., the virtual traffic D (Ns, t) of the virtual node Ns to the user node t is constant and the constraint of the MCF problem is satisfied, item 3. So far, in the embodiment of the present invention, the constraint condition of the MCF problem item 3 can be expressed as
The original network topology G (V, E) (representing the node set of V in the network topology graph, and E representing the link set) is converted into a new network topology G through the above steps*(V*,E*) Later, a new set of nodes V of the network topology*Including node Si, virtual node Ns and router node (e.g., node u of the foregoing embodiment), etc., link set E*Including all physical and virtual links (Ns, Si), the demand matrix in the MCF problem becomes constant. Up to now, it is equivalent to the non-MCF problem in the joint optimization scheme of TE and SS (i.e. in this problem, the traffic demand D(s) of a single server to a single useriT) is variable) into a standard MCF problem. Solving the standard MCF problem is prior art and will not be described herein.
In a new network topology G*(V*,E*) The optimal solution to the MCF problem includes optimal traffic on physical and virtual links (Ns, Si). Virtual link (N)s,si) Represents a server node s for traffic overiUpper total CP traffic, i.e.
And S403, solving a Network Entropy Maximization problem (NEMR) according to the optimal flow on the link to obtain optimal link weights on the virtual link and the physical link.
The NEMR problem refers to a problem expressed using the following expression:
NEMR:
constraint conditions are as follows:
variables are as follows: <math><mrow><msubsup><mi>x</mi><mrow><mi>s</mi><mo>,</mo><mi>t</mi></mrow><mi>j</mi></msubsup><mo>≥</mo><mn>0</mn></mrow></math>
the NEMR problem can be solved using the dual decomposition principle, assuming λu,vThe lagrangian operator introduced for relaxing the constraint condition of item 1 of the NEMR problem, namely the dual variable of the lagrangian dual problem. Let dual variable lambdau,vIs optimally solved asObviously, for virtual link (N)s,si) The flow constraint of (2 nd term constraint), the left half of the inequality is always no greater than the right half thereof. Therefore, the optimal dual variables on all virtual links are enabled by the specially designed NEMR formIs 0.
The solution to the NEMR problem gives the dual variable for each link (including physical and virtual links) in the topology map. The optimal values of these dual variables can be used to calculate how traffic for the same destination is distributed among different neighboring nodes on each node.
Solving the NEMR to obtain the optimal dual variableAndas Link Weights (Link Weights) of the respective links, that is, optimal Link Weights, let w (u, v) be optimal Link Weights, and u and v be two nodes constituting the links. It should be noted that, the optimal dual variables on all virtual linksIs 0, the optimal link weight on the virtual link (Ns, Si) is therefore 0, i.e., w (Ns, Si) is 0.
S404, distributing the optimal link weight to the router nodes on the network topology.
The optimal link flow can be realized by utilizing the optimal set of link weights. Each node may calculate the distribution ratio of traffic on the respective outbound links for the same purpose. Suppose a node s has N outgoing links. The destination node for traffic passing through this node may be any node in the network. For any destination node t, it needs to decide the distribution ratio of the traffic flowing through s on the N outbound links. In a specific implementation, each router of the ISP can independently calculate the assignment ratio for any destination node according to the topology of the entire network and the weight of the link.
Referring to fig. 6, a flow chart of a joint optimization method for traffic engineering and server selection according to another embodiment of the present invention is shown, which mainly includes the following steps:
s601, the server obtains the shortest path length from the router to the user node and the equivalent number from the router to the user node from the router connected with the server.
This step can be implemented in a distributed manner. In particular, the server S of the CPiFrom neighboring router uj(i.e., the router connected to each server of the CP) obtains information including the shortest path length from the router to each customer node tAnd equivalent numberSince each router will calculate and assign the shortest path lengthAnd equivalent numberAnd (5) storing. When the server S of the CPiTo neighboring router ujWhen sending out control information, the router stores the shortest path lengthAnd equivalent numberThe information is transmitted to the server Si. Server SiThat is to sayUsing this information to compute the server SiShortest path length to customer node tAnd equivalent numberWhen all servers get the shortest path lengthAnd equivalent numberThese servers can then exchange these information with each other. After the exchange is finished, each server can independently calculate the corresponding distribution flowAnd then a server selection strategy is made in a distributed mode.
The ISP needs to know the traffic matrix information when performing joint optimization, which includes all background traffic and CP traffic. It is noted that all CP traffic in the traffic matrix starts from a virtual node. The ISP can obtain this information by measurement and evaluation at the underlying network, or can choose to obtain it from the CP.
S602, the server obtains the optimal link weight of each virtual link.
In the embodiment of the present invention, the optimal link weights are obtained by solving the NEMR problem that increases the network topology of the virtual nodes, as described above with reference to the embodiment shown in fig. 4. The virtual link is formed by virtual nodes and servers added in the network topology.
S603, determining the flow demand distributed to the user node on the server according to the obtained optimal link weight, the shortest path length, the equivalent number from the router to the user node t and the total flow demand belonging to the user node.
First, a virtual node Ns is calculated to pass through any virtual link (Ns, S)i) The difference between the shortest path length to the user node t and the shortest path length from the virtual node Ns to the user node t.
The virtual node Ns passes through any virtual link (Ns, S)i) The shortest path length to customer node t is equal to virtual link (Ns, S)i) Is optimized by the optimal link weight w (Ns, S)i) And node SiSum of shortest path lengths (server) to customer node t, i.e.Due to w (Ns, S)i) Is 0, therefore, the virtual node Ns passes through the virtual link (Ns, S)i) The shortest path length to the user node t is actuallyIn practice, the amount of the liquid to be used,wherein, the node ujRepresentation and node Si(Server) neighboring Router, node ujAt the slave SiOn the shortest path to t,represents a node SiTo neighboring node ujIs connected to the link (S)i,uj) Length whenHaving been acquired in S601, the system,it can be easily obtained.
Similarly, the shortest path length from the virtual node Ns to the user node t is adoptedIndicating that similar methods as described above can be usedObtained without repeated explanation.
The shortest path from the virtual node Ns to the user t needs to go through a virtual link, and therefore,and due to virtual links (N)s,Si) Is optimized by the optimal link weight w (Ns, S)i) Is 0, therefore, the virtual node Ns passes through any one virtual link (N)s,Si) The length of the shortest path to the user node t and the shortest path from the virtual node Ns to the user node t (it should be understood by those skilled in the art that the shortest path may also be the shortest path from the virtual node Ns to the user node t through any virtual link (N)s,Si) Shortest path to customer node t) length of the same
Then, according to the equivalent number and difference value from the router to the user node tDetermining total traffic demand on any of the individual virtual links (N)s,Si) The division ratio of (1).
It should be noted that, each virtual link (N)s,Si) Is a server (node S)i) For the total flow demand of the user node t, the total flow demand is any one virtual link (N) in each virtual links,Si) The division ratio of (1), i.e. the total traffic demand needs to be distributed to the corresponding server (node S)i) The flow rate ratio of (c).
As one embodiment of the invention, determining the total traffic demand on any of the individual virtual links (N)s,Si) The above division ratio may be obtained by the following method:
and S1, obtaining the equivalent number from the server to the user node t by the equivalent number from the router to the user node t.
Router (node u)j) Equivalent number to user node tHaving been acquired in step S601, it is not difficult to route the router to the equivalent number acquisition server of the user node t (node S)i) Equivalent number to user node t
S2, calculating the product of the equivalent number from the server to the user node and an attenuation parameter to obtain any virtual link (N)s,Si) The traffic demand split function of.
The decay parameter being the difference in natural index eThe inverse of the power, i.e.,thus, any one virtual link (N)s,Si) The flow demand division function ofTraffic demand split functionWith any of the above virtual links (N)s,Si) And (7) corresponding.
S3, a sum of the traffic division functions corresponding to the respective virtual links is calculated.
The sum of the traffic splitting functions for each virtual link, i.e.
S4, obtaining a flow demand division functionAndthe ratio of (A) to (B) is obtained, and the total required flow is obtained in any virtual link (N)s,Si) The division ratio of (1) or
Finally, the total flow demand and the split ratio are calculatedThe product of (a) to obtain the total flow demand
At any virtual link (N)s,Si) Up-divided flowSince each virtual link (N)s,Si) Is a server (node S)i) For the total flow demand of the user node t, the total flow demand is any one virtual link (N) in each virtual links,Si) The up-divided traffic, i.e. the total traffic demand of the user node t, needs to be allocated to the virtual link (N)s,Si) Corresponding server (node S)i) And (c) the flow rate of (c).
It should be noted that, although the above describes using a distributed method to solve the SS problem, the above description is not limited to the distributed method. In the embodiment of the present invention, the server selection process of the CP may also be handled in a centralized manner. In particular, the server S of the CPiFrom neighboring router ujWhere the shortest path length is obtainedAnd equivalent numberFurther calculates the server SiShortest path length to customer node tAnd equivalent numberIn contrast to the distributed type, the servers SiThe shortest path length from the user node tAnd equivalent numberAnd feeding back to the CP centralized management system. Based on this information, the centralized management system of the CP can make the server selection policy without obtaining the complete topology information from the ISP. This ensures that the ISP and CP obtain an optimal solution while exposing as little mutual information as possible.
As can be seen from the embodiment provided in fig. 6, the CP can independently decide how to distribute the user traffic demand to different servers according to very little information provided by the ISP, thereby solving the SS problem; alternatively, the shortest path length and equivalence number information can be distributively exchanged between servers, and each server can distributively and independently decide its own traffic distribution to different users. In the optimization process, the CP or the server thereof can autonomously update the SS policy of itself only by acquiring a small amount of variable information (for example, shortest path length information and equivalent number information from each router to the user node) from the neighboring router thereof without key information (for example, topology and link state information) of the underlying network. Thus, the PEFT-based TE and SS joint optimization (PETS) is still applicable even in situations where the ISP is unwilling to share underlying network information with the CP.
Referring to fig. 7, a schematic diagram of a logical structure of a router according to an embodiment of the present invention is shown. For convenience of explanation, only portions related to the embodiments of the present invention are shown. The router comprises an outbound link determining module 71, an obtaining module 72 and a splitting policy determining module 73, wherein:
an outbound link determining module 71, configured to determine each outbound link that carries traffic belonging to the same user node;
an obtaining module 72, configured to obtain an optimal link weight of each outbound link determined by the outbound link determining module 71, where the optimal link weight is obtained by solving a network entropy maximization NEMR problem in which a network topology of the virtual node is increased;
and a partitioning policy determining module 73, configured to determine, according to the optimal link weight obtained by the obtaining module 72 and traffic belonging to the same user node, a partitioning policy of the traffic belonging to the same user node on any one of the outbound links (u, v), where u denotes a node where the router is located, and v denotes another node forming the outbound link (u, v).
Further, the partition policy determining module 73 includes a first difference value calculating unit 81, a first partition ratio determining unit 82 and a partition traffic calculating unit 83, as shown in fig. 8, where:
a first difference calculating unit 81, configured to calculate a difference between a shortest path length from a router (node u) to a user node t via any one of the outbound links (u, v) and a shortest path length from the router to the user node t;
it should be noted that the shortest path from the router to the user node t may also be the shortest path from the router (node u) to the user node t via any one of the outbound links (u, v).
A first division ratio determining unit 82, configured to determine, according to the equivalent number from the router to the user node and the difference obtained by the first difference calculating unit 81, a division ratio of the traffic belonging to the same user node on any one of the outward links (u, v);
and a split traffic calculation unit 83, configured to calculate a product of the traffic belonging to the same user node and the split ratio, so as to obtain a split traffic of the traffic belonging to the same user node on any one of the outbound links (u, v).
Further, the first division ratio determining unit 82 includes:
a first product calculation unit for calculating a product of an equivalent number and a decay parameter to obtain a flow division function of any one of the outbound links (u, v), the flow division function corresponding to any one of the outbound links (u, v), the decay parameter being the reciprocal of the power of the difference of the natural exponent e;
the first summation unit is used for calculating the sum of the flow dividing functions corresponding to the outward links;
and the first proportion calculating unit is used for calculating the ratio of the traffic division function to the sum of the traffic division functions corresponding to the outward links to obtain the division proportion of the traffic belonging to the same user node on any outward link (u, v) in the outward links.
Referring to fig. 9, a schematic diagram of a logic structure of a server according to an embodiment of the present invention is provided. For convenience of explanation, only portions related to the embodiments of the present invention are shown. The server comprises a first obtaining module 91, a second obtaining module 92 and a flow determining module 93, wherein:
a first obtaining module 91, configured to obtain, from a router connected to the server, a shortest path length from the router to a user node and an equivalent number from the router to the user node;
a second obtaining module 92, configured to obtain an optimal link weight of each virtual link, where the optimal link weight is obtained by solving a network entropy maximization NEMR problem of a network topology to which a virtual node Ns is added, and the virtual link is formed by the virtual node added in the network topology and the server;
a traffic determining module 93, configured to determine, according to the total traffic demand belonging to the user node, the optimal link weight obtained by the second obtaining module 92, the shortest path length obtained by the first obtaining module 91, and the equivalent number from the router to the user node, a traffic demand allocated to the user node on the server.
Further, the flow rate determining module 93 includes a second difference calculating unit 101, a second division ratio determining unit 102, and a flow rate demand calculating unit 103, wherein:
a second difference calculation unit 101, configured to calculate a difference between a shortest path length from the virtual node to the user node via the arbitrary virtual link (Ns, Si) and a shortest path length from the virtual node to the user node, where Ns represents the virtual node, and Si represents another node constituting the virtual link (Ns, Si);
a second division ratio determining unit 102, configured to determine, according to the difference calculated by the second difference calculating unit 101 and the equivalent number from the router to the user node, a division ratio of the total flow demand on any virtual link (Ns, Si) in the virtual links;
a traffic demand calculation unit 103, configured to calculate a product of the total traffic demand and the partition ratio, so as to obtain a traffic demand, allocated to the user node, of the total traffic demand on a server corresponding to the any virtual link (Ns, Si).
Further, the second division ratio determining unit 102 includes an equivalent number obtaining unit, a second product calculating unit, a second summing unit, and a second ratio obtaining unit, wherein:
an equivalent number obtaining unit, which obtains the equivalent number from the server to the user node according to the equivalent number of the router;
a second product calculation unit, configured to calculate a product of an equivalent number from the server to the user node and a decay parameter, to obtain a traffic demand division function of the any virtual link (Ns, Si), where the traffic demand division function corresponds to the any virtual link (Ns, Si), and the decay parameter is a natural exponent e raised to the power of the difference;
a second summing unit for calculating a sum of the traffic division functions corresponding to the respective virtual links;
and the second proportion calculating unit is used for calculating the ratio of the flow demand division function to the sum to obtain the division proportion of the total flow demand on any virtual link (Ns, Si) in each virtual link.
Referring to fig. 11, a schematic diagram of a logic structure of an optimal link weight obtaining apparatus according to an embodiment of the present invention is shown. For convenience of explanation, only portions related to the embodiments of the present invention are shown. The device comprises a conversion module 111, an optimal flow calculation module 112, an optimal link weight calculation module 113 and a distribution module 114, wherein:
a conversion module 111, configured to convert, by adding a virtual node to a network topology, a traffic demand from each server node to a user node on the network topology into a virtual traffic from the virtual node to the user node, where the virtual traffic is constantly equal to a total traffic demand of the user node, and a virtual link is formed between the virtual node and each server node on the network topology;
an optimal flow solving module 112, configured to use the virtual flow as a constant of a user demand constraint term in a multi-commodity network flow MCF problem, and solve the MCF problem to obtain an optimal flow of each link of the network topology, where each link of the network topology includes a physical link and the virtual link on the network topology, and the user demand constraint term in the MCF problem is:
wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set, respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), and D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand;
an optimal link weight solving module 113, configured to solve the network entropy maximization NEMR problem according to the optimal traffic solved by the optimal traffic solving module 112 on the link, so as to obtain optimal link weights on the virtual link and the physical link;
a distribution module 114 configured to distribute the optimal link weights to router nodes on the network topology.
Referring to fig. 12, a schematic diagram of a logical structure of a joint optimization system for traffic engineering and server selection according to an embodiment of the present invention is provided. For convenience of explanation, only portions related to the embodiments of the present invention are shown. The system includes the optimal link weight obtaining device 121 illustrated in fig. 11, the server 122 illustrated in fig. 9 or fig. 10, and the router 123 illustrated in fig. 7 or fig. 8, wherein:
an optimal link weight obtaining device 121, configured to convert, by adding a virtual node to a network topology, a traffic demand from each server 122 node to a user node on the network topology into a virtual traffic from the virtual node to the user node, where the virtual traffic is equal to a total traffic demand of the user node, a virtual link is formed between the virtual node and each server 122 node on the network topology, and the virtual traffic is a constant of a user demand constraint item in a multi-commodity network flow MCF problem, and the MCF problem is solved to obtain an optimal traffic of each link on the network topology, where each link on the network topology includes a physical link and the virtual link on the network topology, and the user demand constraint item in the MCF problem is:
wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set, respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand, solving a network entropy maximization NEMR problem according to the optimal traffic on the link to obtain optimal link weights on the virtual link and the physical link, and distributing the optimal link weights to a router 123 node on the network topology;
a server 122, configured to obtain, from a router 123 connected to the server, a shortest path length from the router 123 to a user node and an equivalent number from the router 123 to the user node, and obtain an optimal link weight of each virtual link, where the optimal link weight is obtained by solving a network entropy maximization NEMR problem that increases a network topology of virtual nodes, where the virtual links are formed by the virtual nodes added in the network topology and the server, and determine a traffic demand allocated to the user node on the server according to the optimal link weight, the shortest path length, the equivalent number from the router 123 to the user node, and a total traffic demand belonging to the user node;
the router 123 is configured to determine each outbound link carrying traffic belonging to the same user node, obtain an optimal link weight of each outbound link, where the optimal link weight is obtained by solving a network entropy maximization problem NEMR that increases a network topology of a virtual node, and determine, according to the optimal link weight and the traffic belonging to the same user node, a partitioning policy of the traffic belonging to the same user node on any one outbound link (u, v) in each outbound link, where u denotes a node where the router is located, and v denotes another node constituting the outbound link (u, v).
It should be noted that, because the contents of information interaction, execution process, and the like between the modules/units of the apparatus are based on the same concept as the method embodiment of the present invention, the technical effect brought by the contents is the same as the method embodiment of the present invention, and specific contents may refer to the description in the method embodiment of the present invention, and are not described herein again.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable storage medium, and the storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.
The method, the system and the related devices for joint optimization of traffic engineering and server selection provided by the embodiment of the present invention are described in detail above, and a specific example is applied in the present document to explain the principle and the implementation of the present invention, and the description of the above embodiment is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (17)
1. A joint optimization method for traffic engineering and server selection is characterized by comprising the following steps:
the router determines each outward link bearing the flow belonging to the same user node;
the router acquires optimal link weights of the outward links, wherein the optimal link weights are obtained by solving a network entropy maximization NEMR problem of the network topology with the added virtual nodes;
and the router determines a segmentation strategy of the traffic belonging to the same user node on any one of the outward links (u, v) according to the optimal link weight and the traffic belonging to the same user node, wherein u represents a node where the router is located, and v represents another node forming the outward link (u, v).
2. The method according to claim 1, wherein said determining a partitioning policy of the traffic belonging to the same user node on any one of the outbound links (u, v) according to the optimal link weight and the traffic belonging to the same user node comprises:
calculating the difference between the shortest path length from the router to the user node via the any one of the outbound links (u, v) and the shortest path length from the router to the user node;
according to the difference value and the equivalent number from the router to the user node, determining the division proportion of the flow belonging to the same user node on any one outward link (u, v) in each outward link;
and calculating the product of the flow belonging to the same user node and the segmentation proportion to obtain the segmented flow of the flow belonging to the same user node on any one outward link (u, v) in each outward link.
3. The method according to claim 2, wherein said determining the division ratio of the traffic belonging to the same user node on any one of the respective outbound links (u, v) based on the difference and the equivalence number from the router to the user node comprises:
calculating the product of the equivalent number and a decay parameter to obtain a flow segmentation function of any one of the outbound links (u, v), wherein the flow segmentation function corresponds to any one of the outbound links (u, v), and the decay parameter is the reciprocal of the difference power of the natural exponent;
calculating the sum of the flow dividing functions corresponding to the outward links;
and solving the ratio of the flow segmentation function to the sum to obtain the segmentation proportion of the flow belonging to the same user node on any one of the outward links (u, v).
4. The method of claim 2, further comprising:
and the router sends the shortest path length from the router to the user node and the equivalent number from the router to the user node to a server connected with the router.
5. The method of any of claims 1 to 4, wherein the shortest path length from each node in the respective outbound link to the customer node is less than the shortest path length from the router to the customer node.
6. A joint optimization method for traffic engineering and server selection is characterized by comprising the following steps:
the method comprises the steps that a server acquires the shortest path length from a router connected with the server to a user node and the equivalent number from the router to the user node;
the server acquires optimal link weights of all virtual links, wherein the optimal link weights are obtained by solving a network entropy maximization NEMR problem of a network topology with added virtual nodes, and the virtual links are formed by the server with the added virtual nodes in the network topology;
and the server determines the flow demand distributed to the user node on the server according to the optimal link weight, the shortest path length, the equivalent number and the total flow demand belonging to the user node.
7. The method of claim 6, wherein said determining traffic demand allocated to the customer node on the server based on the optimal link weight, the shortest path length, the equivalence number, and a total traffic demand belonging to the customer node comprises:
calculating a difference between a shortest path length from the virtual node to the user node through the arbitrary virtual link (Ns, Si) and a shortest path length from the virtual node to the user node, wherein Ns represents the virtual node, and Si represents another node constituting the virtual link (Ns, Si);
determining the division proportion of the total flow demand on any virtual link (Ns, Si) in each virtual link according to the equivalent number and the difference value;
and calculating the product of the total flow demand and the division ratio to obtain the flow demand of the total flow demand distributed to the user node on the server corresponding to the any virtual link (Ns, Si).
8. The method of claim 7, wherein said determining a split ratio of the total traffic demand over any of the virtual links (Ns, Si) based on the equivalence numbers and the difference comprises:
obtaining the equivalent number from the server to the user node by the equivalent number;
calculating the product of the equivalent number from the server to the user node and a decay parameter to obtain a flow demand division function of the arbitrary virtual link (Ns, Si), wherein the flow demand division function corresponds to the arbitrary virtual link (Ns, Si), and the decay parameter is the reciprocal of the difference power of a natural exponent;
calculating the sum of the flow dividing functions corresponding to each virtual link;
and calculating the ratio of the flow demand division function to the sum to obtain the division proportion of the total flow demand on any virtual link (Ns, Si) in each virtual link.
9. A joint optimization method for traffic engineering and server selection is characterized by comprising the following steps:
the method comprises the steps that a virtual node is added on a network topology, the flow demand from each server node to a user node on the network topology is converted into the virtual flow from the virtual node to the user node, the virtual flow is constantly equal to the total flow demand of the user node, and a virtual link is formed between the virtual node and each server node on the network topology;
the virtual flow is taken as a constant of a user demand constraint item in a multi-item network flow MCF problem, the MCF problem is solved to obtain an optimal flow on each link of the network topology, each link of the network topology comprises a physical link and the virtual link on the network topology, and the user demand constraint item in the MCF problem is as follows:
wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set, respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), and D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand;
solving a network entropy maximization NEMR problem according to the optimal flow on the link to obtain optimal link weights on the virtual link and the physical link;
distributing the optimal link weights to router nodes on the network topology.
10. A router, comprising:
an outward link determining module, configured to determine outward links carrying traffic belonging to the same user node;
an obtaining module, configured to obtain an optimal link weight of each outbound link determined by the outbound link determining module, where the optimal link weight is obtained by solving a network entropy maximization NEMR problem in which a network topology of a virtual node is increased;
and a partitioning policy determining module, configured to determine, according to the optimal link weight and the traffic belonging to the same user node, a partitioning policy of the traffic belonging to the same user node on any one of the outbound links (u, v), where u denotes a node where the router is located, and v denotes another node constituting the outbound link (u, v).
11. The router of claim 10, wherein said split policy determination module comprises:
a first difference calculation unit for calculating a difference between a shortest path length from the router to the user node via the arbitrary one of the outbound links (u, v) and a shortest path length from the router to the user node;
a first division ratio determining unit, configured to determine, according to the equivalent number from the router to the user node and the difference obtained by the first difference calculating unit, a division ratio of the traffic belonging to the same user node on any one of the outward links (u, v);
and the division flow calculation unit is used for calculating the product of the flow belonging to the same user node and the division ratio to obtain the division flow of the flow belonging to the same user node on any one outward link (u, v) in each outward link.
12. The router according to claim 11, wherein the first division ratio determining unit includes:
a first product calculation unit, configured to calculate a product of the equivalent number and a decay parameter to obtain a traffic splitting function of the any one outbound link (u, v), where the traffic splitting function corresponds to the any one outbound link (u, v), and the decay parameter is a reciprocal of the natural exponent to the power of the difference;
a first summing unit for calculating a sum of traffic split functions corresponding to the respective outbound links;
and the first proportion calculating unit is used for calculating the ratio of the flow dividing function to the sum to obtain the dividing proportion of the flow belonging to the same user node on any one of the outward links (u, v).
13. A server, comprising:
the first acquisition module is used for acquiring the shortest path length from the router to a user node and the equivalent number from the router to the user node from the router connected with the server;
a second obtaining module, configured to obtain an optimal link weight of each virtual link, where the optimal link weight is obtained by solving a problem of network entropy maximization NEMR of a network topology to which a virtual node Ns is added, and the virtual link is formed by the virtual node added in the network topology and the server;
and the flow determining module is used for determining the flow demand distributed to the user nodes on the server according to the total flow demand belonging to the user nodes, the optimal link weight obtained by the second obtaining module, the shortest path length obtained by the first obtaining module and the equivalent number from the router to the user nodes.
14. The server of claim 13, wherein the traffic determination module comprises:
a second difference calculation unit configured to calculate a difference between a shortest path length from the virtual node to the user node via the arbitrary virtual link (Ns, Si) and a shortest path length from the virtual node to the user node, where Ns represents the virtual node and Si represents another node constituting the virtual link (Ns, Si);
a second division ratio determining unit, configured to determine, according to the difference calculated by the second difference calculating unit and the equivalent number from the router to the user node, a division ratio of the total flow demand on any virtual link (Ns, Si) in the virtual links;
and the traffic demand calculation unit is used for calculating the product of the total traffic demand and the division ratio to obtain the traffic demand of the total traffic demand, which is distributed to the user node on the server corresponding to the any virtual link (Ns, Si).
15. The server according to claim 14, wherein the second division ratio determining unit includes:
an equivalent number obtaining unit, which obtains the equivalent number from the server to the user node according to the equivalent number of the router;
a second product calculation unit, configured to calculate a product of an equivalent number from the server to the user node and a decay parameter, to obtain a traffic demand division function of the any virtual link (Ns, Si), where the traffic demand division function corresponds to the any virtual link (Ns, Si), and the decay parameter is a reciprocal of the natural exponent e raised to the power of the difference;
a second summing unit for calculating a sum of the traffic division functions corresponding to the respective virtual links;
and the second proportion calculating unit is used for calculating the ratio of the flow demand division function to the sum to obtain the division proportion of the total flow demand on any virtual link (Ns, Si) in each virtual link.
16. An optimal link weight acquisition apparatus, comprising:
the conversion module is used for converting the flow demand from each server node to a user node on the network topology into the virtual flow from the virtual node to the user node by adding the virtual node on the network topology, wherein the virtual flow is constantly equal to the total flow demand of the user node, and a virtual link is formed between the virtual node and each server node on the network topology;
an optimal flow solving module, configured to use the virtual flow as a constant of a user demand constraint term in a multi-commodity network flow MCF problem, and solve the MCF problem to obtain an optimal flow of each link of the network topology, where each link of the network topology includes a physical link and a virtual link on the network topology, and the user demand constraint term in the MCF problem is:
wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set on the network topology G (V, E), respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), and D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand;
the optimal link weight solving module is used for solving a network entropy maximization NEMR problem according to the optimal flow solved by the optimal flow solving module on the link, so as to obtain optimal link weights on the virtual link and the physical link;
a distribution module for distributing the optimal link weight to a router node on the network topology.
17. A combined optimization system for traffic engineering and server selection is characterized by comprising an optimal link weight acquisition device, a server and a router;
the optimal link weight obtaining device is configured to convert, by adding a virtual node to a network topology, a traffic demand from each server node to a user node on the network topology into a virtual traffic from the virtual node to the user node, where the virtual traffic is equal to a total traffic demand of the user node, a virtual link is formed between the virtual node and each server node on the network topology, the virtual traffic is a constant of a user demand constraint item in a multi-commodity network flow MCF problem, the MCF problem is solved to obtain an optimal traffic of each link on the network topology, each link on the network topology includes a physical link and the virtual link on the network topology, and the user demand constraint item in the MCF problem is:wherein (s, V) ∈ E denotes a directed link from the server node s to the node V in the network topology graph G (V, E), (u, s) ∈ E denotes a directed link from the node u to the server node s, V and E denote a node set and a link set, respectively,representing the traffic demand belonging to the user node t on the directed link (s, v),representing the traffic demand belonging to a user node t on a directed link (u, s), D (s, t) representing the traffic demand from a server node s to the user node t, wherein D (s, t) is a constant of a constraint item of the user demand, solving a network entropy maximization NEMR problem according to the optimal traffic on the link to obtain optimal link weights on the virtual link and the physical link, and calculating the optimal link weightsDistributing the optimal link weight to a router node on the network topology;
the server is used for acquiring the shortest path length from the router to a user node and the equivalent number from the router to the user node from the router connected with the server, acquiring the optimal link weight of each virtual link, wherein the optimal link weight is obtained by solving a network entropy maximization NEMR problem of a network topology with the added virtual nodes, the virtual links are formed by the added virtual nodes in the network topology and the server, and the traffic demand distributed to the user node on the server is determined according to the optimal link weight, the shortest path length, the equivalent number from the router to the user node and the total traffic demand belonging to the user node;
the router is configured to determine each outbound link carrying traffic belonging to the same user node, obtain an optimal link weight of each outbound link, where the optimal link weight is obtained by solving a network entropy maximization problem NEMR that increases a network topology of a virtual node, and determine, according to the optimal link weight and the traffic belonging to the same user node, a partitioning policy of the traffic belonging to the same user node on any one outbound link (u, v) in the outbound links, where u denotes a node where the router is located, and v denotes another node constituting the outbound link (u, v).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201010270101.2A CN102136998B (en) | 2010-08-30 | 2010-08-30 | Traffic engineering and server selection joint optimization method, system and related equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201010270101.2A CN102136998B (en) | 2010-08-30 | 2010-08-30 | Traffic engineering and server selection joint optimization method, system and related equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102136998A true CN102136998A (en) | 2011-07-27 |
CN102136998B CN102136998B (en) | 2014-06-04 |
Family
ID=44296658
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201010270101.2A Expired - Fee Related CN102136998B (en) | 2010-08-30 | 2010-08-30 | Traffic engineering and server selection joint optimization method, system and related equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102136998B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102388585A (en) * | 2011-09-19 | 2012-03-21 | 华为技术有限公司 | Network optimization flow control method, device and system |
CN104869044A (en) * | 2015-06-19 | 2015-08-26 | 中国人民解放军信息工程大学 | Mapping method and mapping device for virtual network |
CN104335161B (en) * | 2012-06-18 | 2017-04-12 | 国际商业机器公司 | Efficient evaluation of network robustness with a graph |
CN106571947A (en) * | 2015-11-16 | 2017-04-19 | 中国人民解放军理工大学 | Network element model implementation method supporting complex multi-network construction |
CN107548545A (en) * | 2015-03-30 | 2018-01-05 | 阿尔卡特朗讯公司 | The offline optimization for traffic engineering with Segment routing |
CN111211987A (en) * | 2019-11-29 | 2020-05-29 | 清华大学 | Method and system for dynamically adjusting flow in network, electronic equipment and storage medium |
CN111512600A (en) * | 2017-12-21 | 2020-08-07 | 瑞典爱立信有限公司 | Method, apparatus and computer program for distributing traffic in a telecommunications network |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1688133A (en) * | 2005-04-11 | 2005-10-26 | 西安电子科技大学 | Resource utilization optimization method for transfer network |
WO2009074558A2 (en) * | 2007-12-11 | 2009-06-18 | Thomson Licensing | Device and method for optimizing access to contents by users |
CN101800768A (en) * | 2010-01-19 | 2010-08-11 | 南京邮电大学 | Gridding data transcription generation method based on storage alliance subset partition |
-
2010
- 2010-08-30 CN CN201010270101.2A patent/CN102136998B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1688133A (en) * | 2005-04-11 | 2005-10-26 | 西安电子科技大学 | Resource utilization optimization method for transfer network |
WO2009074558A2 (en) * | 2007-12-11 | 2009-06-18 | Thomson Licensing | Device and method for optimizing access to contents by users |
CN101800768A (en) * | 2010-01-19 | 2010-08-11 | 南京邮电大学 | Gridding data transcription generation method based on storage alliance subset partition |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012149748A1 (en) * | 2011-09-19 | 2012-11-08 | 华为技术有限公司 | Network optimization traffic control method, device and system |
CN102388585B (en) * | 2011-09-19 | 2014-01-08 | 华为技术有限公司 | Network optimization flow control method, device and system |
CN102388585A (en) * | 2011-09-19 | 2012-03-21 | 华为技术有限公司 | Network optimization flow control method, device and system |
CN104335161B (en) * | 2012-06-18 | 2017-04-12 | 国际商业机器公司 | Efficient evaluation of network robustness with a graph |
CN107548545A (en) * | 2015-03-30 | 2018-01-05 | 阿尔卡特朗讯公司 | The offline optimization for traffic engineering with Segment routing |
CN104869044A (en) * | 2015-06-19 | 2015-08-26 | 中国人民解放军信息工程大学 | Mapping method and mapping device for virtual network |
CN104869044B (en) * | 2015-06-19 | 2019-01-29 | 中国人民解放军信息工程大学 | A kind of virtual net mapping method and device |
CN106571947A (en) * | 2015-11-16 | 2017-04-19 | 中国人民解放军理工大学 | Network element model implementation method supporting complex multi-network construction |
CN106571947B (en) * | 2015-11-16 | 2019-08-23 | 中国人民解放军理工大学 | A kind of network element model implementation method for supporting complicated pluralism net structure |
CN111512600A (en) * | 2017-12-21 | 2020-08-07 | 瑞典爱立信有限公司 | Method, apparatus and computer program for distributing traffic in a telecommunications network |
CN111512600B (en) * | 2017-12-21 | 2022-04-29 | 瑞典爱立信有限公司 | Method and apparatus for distributing traffic in a telecommunications network |
CN111211987A (en) * | 2019-11-29 | 2020-05-29 | 清华大学 | Method and system for dynamically adjusting flow in network, electronic equipment and storage medium |
CN111211987B (en) * | 2019-11-29 | 2021-07-27 | 清华大学 | Method and system for dynamically adjusting flow in network, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN102136998B (en) | 2014-06-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102136998B (en) | Traffic engineering and server selection joint optimization method, system and related equipment | |
Li et al. | On social-aware content caching for D2D-enabled cellular networks with matching theory | |
Key et al. | Path selection and multipath congestion control | |
JP3115769B2 (en) | How to assign metric values to each link in the network | |
Hussain et al. | Genetic algorithm based adaptive offloading for improving IoT device communication efficiency | |
Ghosh et al. | Scalable multi-class traffic management in data center backbone networks | |
Feng et al. | Postcard: Minimizing costs on inter-datacenter traffic with store-and-forward | |
EP2515478B1 (en) | Method, apparatus and system for joint optimizations | |
KR20060022680A (en) | Method and system for global routing and bandwidth sharing | |
Zhang et al. | Mapping multicast service-oriented virtual networks with delay and delay variation constraints | |
CN103001892A (en) | Network resource distribution method and system based on cloud computing | |
Hou et al. | Bandwidth scheduling for big data transfer using multiple fixed node-disjoint paths | |
Papadaki et al. | Gateway selection and routing in wireless mesh networks | |
Wong | Telecommunications network design: Technology impacts and future directions | |
Lu et al. | A layered coalitional game framework of wireless relay network | |
Iosifidis et al. | Distributed storage control algorithms for dynamic networks | |
Kashyap et al. | Routing and traffic engineering in hybrid RF/FSO networks | |
Antoniou et al. | Access network synthesis game in next generation networks | |
Huang et al. | Efficiency and Braess' paradox under pricing in general networks | |
Zheng et al. | Optimal peer-to-peer technique for massive content distribution | |
CN101977159A (en) | Management method of bandwidth resources of narrow band network | |
Lin et al. | Scheduling algorithms for time-constrained big-file transfers in the Internet of Vehicles | |
Fortz | Applications of meta‐heuristics to traffic engineering in IP networks | |
Muscariello et al. | Do next generation networks need path diversity? | |
Liu et al. | Online Virtual Network Embedding for Both the Delay Sensitive and Tolerant Services in SDN-enabled Satellite-Terrestrial Networks |
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: 20140604 Termination date: 20190830 |
|
CF01 | Termination of patent right due to non-payment of annual fee |