CN105743804A - Data flow control method and system - Google Patents

Abstract

The invention relates to the communication field, and discloses a data flow control method and system. The method includes: calculating the data flow table request amount of each switch node in the software-defined network SDN; optimizing the weight of each link in the SDN according to the aggregation cost of each switch node and the data flow table request amount of each switch node, To balance the data flow of each switch node. The invention realizes the load balance between the switches in the network by optimizing the route of the network data flow, fully utilizes the limited flow table processing capability and the flow table cache space, and improves the utilization efficiency of network resources.

Description

A data flow control method and system

technical field

The present invention relates to the network field, in particular to a data flow control method and system.

Background technique

In the software-defined network SDN, the switching device performs high-speed data forwarding according to the flow table on the data plane, and the controller is responsible for routing calculation, flow table generation and maintenance on the control plane, so that network resources can be allocated more reasonably. SDN aims to realize the definition and open interface of network interconnection and network behavior, so as to support the innovation of various new network architectures and new services in the future. Different from the thin-waisted architecture of the original TCP/IP, SDN migrates the control function from the traditional distributed network equipment to the controllable computing equipment, and finally realizes the automatic control function of the network through the open and programmable software mode.

In an SDN network, when a data flow arriving at a switch has no matching flow entry, the first data packet of the data flow will be sent to the controller. The controller calculates the forwarding path and sends it to the corresponding switch. It takes a certain amount of time for the switch to install the forwarding rules. When the amount of data is large, due to the limited installation speed of the flow table of the switch and the limited processing capacity of the flow table of the controller, bottlenecks may occur on both the switch and the controller.

In order to solve this problem, the following technical solutions are provided in the prior art:

By using clustering or fuzzy matching to classify data streams, the controller can simultaneously process a class of data streams instead of a single stream to reduce the burden. The looser the data flow matching rules, such as using a mask with fewer bits, the fewer types are divided, and the lighter the burden on the controller. However, when the technical solution uses a shorter subnet mask for data flow matching, the granularity of data flow in the network increases. The accuracy of data flow control by network administrators is reduced, and it is difficult to perform precise data filtering, access control and other operations. At the same time, it also increases the allocation granularity of resources such as links and storage space in the network, which cannot realize efficient utilization of network resources.

Dynamic flow shaping policies that focus on "elephant flows" through the switch. The controller estimates data transfer requests for flows exceeding a certain rate, calculates the optimal path, and installs flow-specific forwarding flow entries. This technology requires the switch to collect statistics on each flow in real time to detect and analyze whether it conforms to the characteristics of "elephant flow". If "elephant flow" is found, the controller will be notified for targeted processing. In the case of heavy network load, these real-time monitoring and additional computing work will further increase the burden on switches and controllers, which is not conducive to improving network performance.

Contents of the invention

The purpose of the present invention is to provide a data flow control method and system to balance the data flow of each switch in the SDN.

In order to achieve the above object, the present invention provides a data flow control method, the method comprising: calculating the data flow table request amount of each switch node in the software-defined network SDN; according to the aggregation cost of each switch node and the data of each switch node The flow table request volume optimizes the weight of each link in the SDN to balance the data flow of each switch node; the aggregation cost of the nth switch $\mathrm{\Φ}\left(Load\right(\mathrm{no}\left)\right)=\underset{i}{\Σ}\mathrm{\Φ}\left({\mathrm{match}}_i\right)\×{\mathrm{load}}_i\left(\mathrm{no}\right),$ Φ(Match _i )＝γ _b Conv(Match _i ), Conv(Match _i )＝F _up (Match _b )/F _up (Match _i ), Load(n) is the data forwarding rate of the nth switch, Load _i (n) is the data forwarding rate of the nth switch using the data flow table matching method Match _i , F _up (Match _b ) is the data flow table request amount using the reference data flow table matching method Match _b , F _up (Match _i ) is the request amount of the data flow table using the matching mode Match _i , n is an integer, i is an integer, and γ _b is the convergence cost of forwarding 1-bit data by the preset switch.

Preferably, Load _i (n) is calculated according to the following formula:

Among them, Match _P is the most accurate data flow table matching method, Match _K is the most relaxed data flow table matching method, P is an integer, K is an integer, K is greater than or equal to P+1, and F _up (n) is the nth The upper limit of the data flow table request amount of the switch, F _up (i) is the flow table request amount within the timeout time T of the data flow table, F _i (n) is the number of data flows forwarded by the nth switch within the timeout time T , F _size is the average size of a single data stream.

Preferably, the data flow table request amount of the nth switch node Where α _b is the ratio of the data flow table request volume of the nth switch using the benchmark matching method Match _b to the total number of data flows forwarded by the switch per unit time.

Preferably, optimizing the weight of each link in the SDN according to the aggregation cost of each switch node and the data flow table request amount of each switch node includes: The request amount of the data flow table is less than or equal to the upper limit of the request amount of the data flow table of each switch node UB is a boundary condition to optimize the weight of each link in the SDN.

Preferably, optimizing the weight of each link in the SDN according to the aggregation cost of each switch node and the data flow table request amount of each switch node includes optimizing the weight of each link in the SDN according to the following formula:

$<span\; class="notranslate">\; \{</span>\begin{array}{c}\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\\ \underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; B</span>}\end{array}<span\; class="notranslate">\; .</span>$

Preferably, optimizing the weight of each link in the SDN according to the aggregation cost of each switch node and the request amount of the data flow table of each switch node includes optimizing the weight of each link in the SDN by solving the shortest path through a genetic algorithm .

Correspondingly, the present invention provides a data flow control system, the system includes each switch node in the software-defined network SDN and a network controller, and the network controller includes: a calculation module, used to calculate each switch node in the software-defined network SDN The amount of data flow table requests; the balance module is used to optimize the weight of each link in the SDN according to the aggregation cost of each switch node and the data flow table request amount of each switch node, so as to balance the data flow of each switch node ; where the aggregation cost of the nth switch $\mathrm{\&Phi;}\left(Load\right(\mathrm{no}\left)\right)=\underset{i}{\&Sigma;}\mathrm{\&Phi;}\left({\mathrm{match}}_i\right)\&times;{\mathrm{load}}_i\left(\mathrm{no}\right),$ Φ(Match _i )＝γ _b Conv(Match _i ), Conv(Match _i )＝F _up (Match _b )/F _up (Match _i ), Load(n) is the data forwarding rate of the nth switch, Load _i (n) is the data forwarding rate of the nth switch using the data flow table matching method Match _i , F _up (Match _b ) is the data flow table request amount using the reference data flow table matching method Match _b , F _up (Match _i ) is the request amount of the data flow table using the matching mode Match _i , n is an integer, i is an integer, and γ _b is the convergence cost of forwarding 1-bit data by the preset switch.

Preferably, the calculation module calculates Load _i (n) according to the following formula:

Among them, Match _P is the most accurate data flow table matching method, Match _K is the most relaxed data flow table matching method, P is an integer, K is an integer, K is greater than or equal to P+1, and F _up (n) is the nth The upper limit of the data flow table request amount of the switch, F _up (i) is the flow table request amount within the timeout time T of the data flow table, F _i (n) is the number of data flows forwarded by the nth switch within the timeout time T , F _size is the average size of a single data stream.

Preferably, the data flow table request amount of the nth switch node Where α _b is the ratio of the data flow table request volume of the nth switch using the benchmark matching method Match _b to the total number of data flows forwarded by the switch per unit time.

Preferably, the balance module is also used to optimize the SDN according to the aggregation cost of each switch node and with the data flow table request amount of each switch node being less than or equal to the upper limit UB of the data flow table request amount of each switch node as a boundary condition The weight of each link in .

Preferably, the balance module is also used to optimize the weight of each link in the SDN according to the following formula:

$<span\; class="notranslate">\; \{</span>\begin{array}{c}\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\\ \underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; B</span>}\end{array}<span\; class="notranslate">\; .</span>$

Preferably, the balance module is also used to optimize the weight of each link in the SDN by solving the shortest path through a genetic algorithm.

This method starts from the perspective of load balancing between switches and controllers in SDN, takes node performance as the main limiting factor, and performs traffic engineering through link weight adjustment. The present invention realizes the load balance between the switches in the network by optimizing the route of the network data stream, fully utilizes the limited flow table processing capability and the flow table cache space, improves the utilization efficiency of network resources, and has burst traffic in the network, etc. In heavy load scenarios, avoid or reduce the overload of switches and controllers, improve network throughput, reduce end-to-end delay of data packets, and optimize network performance.

Other features and advantages of the present invention will be described in detail in the detailed description that follows.

Description of drawings

The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the description, together with the following specific embodiments, are used to explain the present invention, but do not constitute a limitation to the present invention. In the attached picture:

Figure 1 is a schematic diagram of traffic engineering in SDN;

Fig. 2 is a flow chart of genetic algorithm solving routing optimization problem;

Fig. 3 is a schematic diagram of data flow table request volume and average matching accuracy;

Fig. 4 is a schematic diagram of the relationship between the cost function and the load;

Fig. 5 is a schematic diagram of a network controller.

Explanation of reference signs

100 calculation modules 200 equalization modules

detailed description

Specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described here are only used to illustrate and explain the present invention, and are not intended to limit the present invention.

In the SDN network, when data arrives at the switch, the switch will look up the routing table entries in its data flow table cache. If there is a matching data flow entry, take the corresponding action, such as forwarding to a specific port. If there is no match, the first packet is sent to the controller. The controller calculates the forwarding path according to the global view, and delivers it to all the switches involved in the path. For the data flow or the subsequent data packets of the same type of data flow, each switch in the forwarding path will forward according to the data flow entries installed therein. The data flow entry in the data flow table has a certain lifetime, and the data flow entry that expires will be deleted.

When processing according to the data flow table, the data packet is matched and selected with the data flow entries in the data flow table. Among them, the matching field (MatchFields) can be extracted from the data packet, including source MAC address, destination MAC address, Vlanid, Vlan priority, source IP address, destination IP address, TCP/UDP source port number and destination port number, etc. 12 domains. Each field can be a certain value or arbitrary, or an exact match can be achieved through a mask. The precise matching of the data flow tables on layers 2-4 makes network management more simplified and flexible.

However, due to the limited performance of controllers and switches, they are also facing greater challenges. For example, when the number of switches increases, the amount of data flow table requests sent to the controller will increase exponentially. Currently, the NOX controller's ability to process packet-in messages is 30K/s. Centralized control by a single controller may cause a performance bottleneck in the system. This limits the amount of data flow table requests sent by the switch to the controller, generally no more than 60-330 per second. Under a certain limit on the number of data flow table updates, the larger the amount of data forwarded by the switch, the more data flow table requests are required. In order not to exceed the performance limit of the controller, it is necessary to reduce the amount of data flow table requests.

In addition, for the switch, the installation of each data flow table requires about 10ms delay, and the performance limit of updating data flow entries is 800-1000 entries/second. The data flow table in the switch is divided into a linear table and a hash table, where the linear table is searched according to wildcards, the traversal method, the time complexity is O(n), and the hash table is matched according to the determined matching value, that is, the data flow 12-tuple in the table, the time complexity is O(1). Since the OpenFlow device needs to support this data flow table forwarding method of tuple lookup, a network chip that supports this architecture is also required. Although the existing TCAM technology can support this feature, the price of a large-capacity TCAM is extremely expensive, so the data flow table space is limited. The data flow table capacity of a linear table is generally 100, and the data flow table capacity of a single hash table is 8K, 64K or 128K. In practical applications, it is difficult to deal with flexible and large-scale network traffic.

For the SDN network, if the routing planning of the data flow is not reasonable enough, as the amount of data requests in the network increases, it is inevitable that some controllers will be overloaded. At this time, the data flow table may not be requested and updated in time. It causes network congestion, increased data transmission delay, and decreased network throughput and QoS. If the problem is solved only by reducing the matching accuracy of the data flow table, it will affect the flow control performance of the network. As a result, a large number of switches are still under light load, resulting in a waste of network resources such as forwarding capability and buffer space. As shown in Figure 1, there are switches A-E. In (a) of Figure 1, due to the excessive data requests from A->B->C, switch A, switch B and switch C are overloaded. In (b) of Figure 1, through traffic engineering, the route of some traffic is changed to A->D->E->C to achieve load balancing between switches.

Data flow table matching methods with different accuracy will significantly affect the amount of data flow table requests and data flow table space usage of the switch. The present invention firstly defines the matching accuracy of the data flow entry. The matching fields of the data flow entry may include a total of 12 fields such as Ethernet source and destination addresses, IP source and destination addresses, and the like. In the present invention, three matching items of IP source address (src), IP destination address (dst) and destination port number (port) are mainly considered. Different subnet masks combine to produce different matching accuracy. As shown in the table below, define the following four data flow table matching methods and their matching accuracy.

Table Data Flow Table Matching Mode and Accuracy Table

When the number of network segments or hosts connected to the switch increases, the network traffic fluctuates greatly, or when the network is initialized, the number and types of data flows in the network increase, and the probability that the data flow cannot match the cached data flow table in the switch increases. Send a large number of packet-in messages to request the data flow table from the controller. Due to the limited processing capability of the controller for packet-in, there is an upper limit for the number of data flow table requests F _up of all connected switches, and F _up ≤ UB.

As the amount of data forwarded by the switch increases, more data flow table requests are required. In order not to exceed the limit on the number of data flow table requests, a matching method with a shorter mask length must be used. At this time, the controller's ability to control the data flow in the network and the flexibility of management are reduced. In order to reflect the benefits brought by exact matching, the aggregation cost of the defined flow is proportional to the aggregation ability of the matching mode for the flow.

The baseline matching method can be determined according to the traffic condition in the network and the performance of the switch. For example, you can select /32src+/32dst+port as the base matching mode of the access switch, and /24dst+/24src+port as the base matching mode of the core switch. The benchmark matching method can be understood as the most accurate matching degree that all data flows can achieve when the traffic is the smallest according to the switch traffic statistics.

When the average rate of the forwarding data flow of the switch is F data flows per second, the data flow table request volume when the benchmark matching method Match _b is adopted is F _up (Match _b )=α _b F, and it is defined that the switch forwards 1 bit at this time The aggregation cost of data is γ _b , and the aggregation ability of different matching modes Match _i is

Conv(Match _i )＝F _up (Match _b )/F _up (Match _i )

The aggregation cost when using Match _i to forward 1-bit data is

Φ(Match _i )＝γ _b Conv(Match _i )

Assuming that there are N switch nodes in the network, the total data forwarding rate of the nth switch is Load(n)(Mbps), and the amount of data forwarded by Match _i is Load _i (n)(Mbps), then In the following, given Load(n), the matching scheme Load _i (n) is calculated.

When the loosest matching method of the data flow is Match _K , the maximum amount of data that the switch can forward is MaxLoad _K. In order to achieve as precise a match as possible, Match _K should satisfy a minimum of the following conditions.

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; MaxLoad</span>}}_{\mathrm{<span\; class="notranslate">\; K</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\\ {\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; B</span>}\end{array}\right.$

Assume that the most accurate match that the data flow forwarded by the switch can achieve is Match _P . Load _i (n) is the data traffic of switch n using the i-th matching method; MaxLoad _i is the maximum traffic load that the switch can handle when the i-th matching method is adopted and the flow table request volume does not exceed the limit; i=1, 2, ..., K.

(1) Examples of solutions for MaxLoad _i and P

It is assumed that the traffic monitoring data of switches in the network is known. During the monitoring process, when the timeout period of the flow table is set as T=10s, the proportional relationship between the request amount of the flow table and the amount of data forwarded by the switch is analyzed below.

It is known that the total amount of data forwarded by the switch per unit time is Load(n)=689.45Mbps, then the total amount of data forwarded in 10 seconds is

Load _T (n)＝Load(n)*T＝689.45Mbit/s*10s＝6.8945Gbit

Knowing that the average size of a single data flow is F _size = 0.16Mbit, the total amount of data flow forwarded in 10S can be calculated, that is, F _T (n) = Load _T (n)/F _size = 6.8945Gbit/0.16Mbit = 43.09K indivual. Wherein, F _i (n) is the number of data streams forwarded by the nth switch within the timeout period T, and F _size is the average size of a single data stream.

For each matching method, the flow table request volume F _up (i) of 10S is known, and the ratio P _up (i)=F _up (i)/F _10s of the flow table request volume to F _10s is calculated as follows:

Type Match _i Match ₁ Match ₂ Match ₃ Match ₄ F _up (i) (articles/second) 800 500 250 220 P _up (i) 1.86% 1.16% 0.58% 0.51%

For a certain switch, assuming that all Match _i are used, the maximum throughput MaxLoad _i (Mbps) that can be accommodated is calculated. According to the limitation of update performance of switch flow table F _up ≤330, then

max(F _up )=(MaxLoad _i ×T/F _size )×P _up (i)=330

Solutions have to:

MaxLoad ₁ = 283.8Mbps

MaxLoad ₂ = 455Mbps

MaxLoad ₃ = 910Mbps

MaxLoad ₄ = 1035Mbps

For example, it is known that the total data forwarding rate of all ports of a certain exchange is Load(n)=800Mbps, then according to

MaxLoad ₂ <Load(n)≤MaxLoad ₃ , the solution is P=2.

In the case of matching as precisely as possible, Match ₂ matching (P=2) is used first, and Match ₃ matching (K=3) is used for the rest.

(2) Load _i (n) solution example:

Assuming that its total throughput is Load(n)(Mbps), and the amount of data forwarded by the matching method Match _i is Load _i (n), then

Load(n)＝Load ₂ (n)+Load ₃ (n)＝800

The total number of data streams processed by the switch 10S is

F _10s (n) = Load (n) × 10/0.16 = 50000

Among them, the number of data streams processed by Match _i is

F _i (n) = Load _i (n) × 10/0.16

Then the total flow table update request amount is

$\begin{array}{c}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\overset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 3</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>\\ <span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>\\ <span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 1.16</span>}\mathrm{<span\; class="notranslate">\; \%</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.58</span>}\mathrm{<span\; class="notranslate">\; \%</span>}\end{array}$

According to F _up (n)≤330, the following equations are obtained

Solutions have to:

$\left\{\begin{array}{c}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 6810</span>}\\ {\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 43190</span>}\end{array}\right.$

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 108.96</span>}\\ \mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 691.04</span>}\end{array}\right.$

At this time, F _up (n) = 330

(3) Summarize the solutions of Load _i (n) and P

To abstract the above example, calculate Load _i (n) according to the following equations:

Among them, Match _P is the most accurate data flow table matching method, Match _K is the most loose data flow table matching method, P is an integer, and K is an integer.

When the Load(n) of the switch does not exceed MaxLoad _P+1 , match as many data streams as possible using Match _P , and match the rest using Match _P+1 . That is, P is the smallest positive integer that satisfies the equation MaxLoad _P <Load(n)≤MaxLoad _P+1 . In the case of matching as precisely as possible, K=min(P+1,4); in the case of the lowest requirement for matching accuracy, K=4.

The data flow table request volume of the switch node is estimated as:

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>$

At this point, the average matching accuracy of the defined switch n∈N is:

$\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; Accu</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>$

Define the average matching accuracy of the network as:

$\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>$

The present invention will optimize the route by setting the link weight in the SDN, improve the network performance, and realize the load balance of the switch. The optimization problem model of OSPF weight setting in SDN is established, and the optimization objective function of the system is obtained.

Suppose an SDN network is represented by G=(N,A), where n∈N represents a switch in the SDN network, and a∈A represents a link in the SDN network. The flow request matrix D is an N×N matrix, and for each node pair (s,t)∈N×N, d _st =D(s,t) represents the flow request between source and destination nodes. A data flow table generated by the controller determines how d _st completes the data transmission through the switches in the network.

Traffic engineering is to seek a routing rule based on a specific objective function, and map the traffic request D to each link in the network, so that the objective function obtains the optimal value. For example, the shortest path algorithm Dijkstra and Bellman-Ford with the path cost as the objective function. These algorithms are simple to operate and widely used.

The link weight setting problem is to assign a non-negative integer weight w _a ∈ [1, w _max ] to each edge a ∈ A. For a given link weight setting w and traffic Demand matrix D, the total data forwarding load Load(n) and matching scheme Load _i (n) of each switch can be calculated.

According to the aggregation cost Φ(Match _i ), the cost function of the switch is defined as

$\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>$

Then the link weight setting problem can be expressed as: find the optimal setting scheme w ^* ＝<w ₁ ^* ,w ₂ ^* ,...,w _A ^* >, where w _i ^* means link i∈1,2 ,...the optimal weight of A, Make Φ _OSPF(w) the smallest. That is, the optimization objective function of the system is:

$\begin{array}{c}{\mathrm{<span\; class="notranslate">\; \&Phi;</span>}}_{\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Phi;</span>}}_{\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; w</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; min</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; load</span>}}^{\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; e</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; load</span>}}^{\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; e</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\end{array}$

where the boundary conditions are $f_{up}=\underset{i\&le;K}{\&Sigma;}{f}_{up}\left({\mathrm{match}}_i\right)=\underset{i\&le;K}{\&Sigma;}{\mathrm{\&alpha;}}_b{\mathrm{load}}_i\left(\mathrm{no}\right)/Co\mathrm{no}v\left({\mathrm{match}}_i\right)\&le;uB.$

(1) Access Switch (AccessSwitch)

The switch is at the edge of the network and is responsible for connecting end hosts to the network. The amount of data forwarded is average, and the link bandwidth of the connection is not high. Due to the connection of a large number of end users, the access switch handles many types of traffic, and the source address, destination address, port number, etc. of the data flow often change. It is more frequent that the switch cannot match the existing flow entries. Due to the large number of access switches, in order to reduce the burden on the controller, the boundary conditions are set as:

F _up ≤UB＝330

(2) Core Switch (CoreSwitch)

The switch is located in the center of the network and is responsible for the forwarding of a large amount of data, and the link bandwidth of the connection is high. Since the number of core switches is small, they mainly process aggregated data flows, and the types and changes of traffic are few, so the flow table requests are small, and the boundary conditions are changed to consider the delay caused by the flow table issued by the installation controller. According to the switch The installation delay of the flow entry, the boundary conditions are set as:

F _up ≤UB＝1000

The data sources for the above boundary conditions are the performance indicators of the switch. The flow table request rate of the SDN switch is about 60-330 per second, and the installation delay of the flow table item is about 10 milliseconds. In specific applications, corresponding adjustments can be made according to actual parameters.

In addition, the number of switches N _Access in the access network and the number N _Core of switches in the core network and the corresponding cost functions are distinguished.

In order to optimize the above-mentioned objective function, a genetic algorithm is used in the present invention, as shown in FIG. 2 , and the specific description is as follows.

In the feasible domain, each individual is coded and represented by a different gene. For a certain generation of population, the genetic code of the individual is mapped to the fitness function value. Then according to a certain selection strategy, select individuals whose fitness value meets the requirements. Through inheritance, mutation and selection, it gradually approaches the optimal solution of the optimization problem.

Assuming that P(t) is used to represent the population of generation t, the main steps of the algorithm include:

(1) Create an initial population P(0) consisting of several individuals.

(2) In the current population, the following steps are repeated until the maximum genetic algebra is reached.

a. Use the cost function to calculate the fitness function value for each individual.

b. Select parent individuals from P(t).

c. Use the selected parent individuals to form the offspring population P(t+1) through crossover and mutation operations.

d. Calculate the fitness function value of each individual in the offspring population, and select the optimal solution of this iterative process.

The following discussion mainly focuses on access switches, and does not involve core switches. The access switch is mainly responsible for connecting users to the network. The number of access switches connected to the controller is relatively large, for example, a dozen to dozens. Considering the limitation of the ability of the controller to process packet-in messages, it is assumed that F _up of a single switch ≤ UB=330 packets per second. According to the measured data of the switch, set Match _b =/32src+/32dst+port, γ _b =1, and the aggregation cost in different matching modes is as follows. Technicians can optimize and adjust the above parameters according to the actual traffic statistics of network operation to achieve the best network performance.

Access switch matching mode aggregation cost table

The performance analysis of the access switch is shown in Figure 3.

As shown in (a) in FIG. 3 , when the load of the access switch does not exceed 283.8 Mbps, the request volume of the data flow table increases as the load increases. All data flow tables use /32src+/32dst+port matching, and the average matching accuracy is 1. As shown in (b) in Figure 3, after the load exceeds 283.8 Mbps, the data flow table request volume remains at 330 per second by adjusting the matching method of the data flow. As the matching method gradually becomes blurred, the matching accuracy gradually decreases. When the load exceeds 1035Mbps, all data flow tables use /24dst matching, and the average matching accuracy is 0.3.

According to this, the piecewise linear functional relationship between the cost function of the switch and the load is obtained as shown in Figure 4, where

$\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>& <span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; 283.8</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\\ \mathrm{<span\; class="notranslate">\; 2.6</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 454.08</span>}& <span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 283.8</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; 455</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\\ \mathrm{<span\; class="notranslate">\; 4.8</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1455.08</span>}& <span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 455</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; 910</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\\ \mathrm{<span\; class="notranslate">\; 6.5</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3002.08</span>}& <span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 910</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; 1035</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\\ \mathrm{<span\; class="notranslate">\; 100</span>}\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 99774</span>}& <span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ></span>\mathrm{<span\; class="notranslate">\; 1035</span>}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\end{array}\right.<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; N</span>}}^{\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}}$

Next, use the genetic algorithm to solve the optimal weight of the link.

(1) Chromosomal coding. In order to find the optimal OSPF link weight setting, the optimal solution can be expressed as a point in the discrete search domain [1, w _max ] ^A. Use w=<w ₁ ,w ₂ ,...,w _A > to represent a weight setting scheme, where w _i represents the weight of link i∈1,2,...A. A point in the search space represents a feasible solution. In order to reduce the calculation amount of the algorithm, 65535 is not used as the upper limit of the weight, but w _max =20 is used here.

(2) Create an initial population. Set the initial population size NIND = 100. Among them, the first 98 individuals are random points in the search domain [1,20] ^A , represented by a vector composed of integer elements. The link weights of the 99th individual are all set to 1. The 100th individual sets the link weight in a manner that is inversely proportional to the link bandwidth, for example, w _i =1G/c _a . When c _a =100M, w _i =10; when c _a =1000M, w _i =1.

(3) Individual fitness evaluation. For each feasible solution, calculate its fitness function value. The specific steps are:

A. Calculate the shortest path according to the Floyd algorithm, and determine the load of each switch according to the traffic request in the network. The specific steps to calculate the load Load(n) of the switch n∈N are as follows:

foralld _st > 0do//Calculate the routing path_st between s and t according to floyd_Path

forallq=1,2,...,length(path_st)-1do

m ₁ ＝path_st(q)m ₂ ＝path_st(q+1)//Two endpoints of the qth path

Load(m ₁ )＝Load(m ₁ )+d _st //Calculate the Load of the first endpoint

Load(m ₂ )＝Load(m ₂ )+d _st //Calculate the Load of the second endpoint

end

end

B. Calculate the cost function value

${\mathrm{<span\; class="notranslate">\; \&Phi;</span>}}_{\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; w</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; load</span>}}^{\mathrm{<span\; class="notranslate">\; A</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; e</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; e</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; load</span>}}^{\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; e</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

C. Calculate the fitness function value

According to the individual's Φ _OSPF(w) value linearly sorted from large to small, assuming that the position of an individual in the population is Pos, the population size is Nind, then its fitness value is

$\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}$

D. Parent selection. Select parent individuals from P(t). Selection should make individuals with large fitness function values more likely to participate in reproduction. At the same time, the selection has a certain degree of randomness to avoid local optima caused by local greed. Random traversal sampling is used here, and the probability of each individual entering the next generation is equal to the ratio of its fitness value to the sum of individual fitness values in the entire population.

E. Form the offspring population. Using the selected parent individuals, the offspring population P(t+1) is formed through crossover and mutation operations. Crossover: For a randomly selected pair of parent individuals, the discrete recombination is used to perform the crossover operation. For each gene, the value of the variable contributed to the offspring was randomly picked by the parent with equal probability. Mutation: For a real-valued population, use the given probability to mutate each variable and return the mutated new population.

F. Carry out (C)-(E) in a loop, and when the reproduction of the population reaches the maximum genetic algebra, the genetic algorithm ends. Record the optimal link weight setting scheme at this time w ^* =<w ₁ ^* ,w ₂ ^* ,...,w _A ^* >.

After obtaining the optimal link weight, the data flow in the network is routed according to the shortest path first. When the weight of each link is determined, the shortest path between all source nodes and destination nodes can be calculated by using the Floyd algorithm. That is, the path with the smallest sum of weights of each link. When the combination of link weights changes, the calculated shortest path changes accordingly, and the route of data flow in the network also changes accordingly, realizing traffic engineering. The following briefly introduces the Floyd algorithm.

Let the interface of the algorithm be floyd(d,n,s,t), where d is the adjacency matrix, where, if switch i is connected to switch j (i≠j), then d(i,j) is the distance between i and j The weight of the link; if not connected, then d(i,j)=∞; when i=j, d(i,j)=0. n is the number of switch nodes, s and t are the dpids of the source and destination switches respectively. Use the floyd algorithm to continuously traverse the nodes, update the shortest path, and save the path cost and each hop route in the cost and route matrices, where the cost matrix is equal to the d matrix when it is initialized.

Given the switch numbers of the source and destination nodes, the corresponding route and cost are returned. Wherein, route is a string composed of a node number and a ">" symbol, for example, "1>2>5" indicates that the path from source node 1 to destination node 5 is 1-2-5. )

Correspondingly, the present invention provides a data flow control system, which includes each switch node and a network controller in a software-defined network (SDN). As shown in Figure 5, the network controller includes: a calculation module, used to calculate the data flow table request volume of each switch node in the software-defined network SDN; The data flow table request volume of the node optimizes the weight of each link in the SDN to balance the data flow of each switch node; the aggregation cost $\mathrm{\&Phi;}\left(Load\right(\mathrm{no}\left)\right)=\underset{i}{\&Sigma;}\mathrm{\&Phi;}\left({\mathrm{match}}_i\right)\&times;{\mathrm{load}}_i\left(\mathrm{no}\right),$ Φ(Match _i )＝γ _b Conv(Match _i ), Conv(Match _i )＝F _up (Match _b )/F _up (Match _i ), Load(n) is the data forwarding rate of the nth switch, Load _i (n) is the amount of data forwarded using the data flow table matching method Match _i , F _up (Match _b ) is the data flow table request amount using the reference data flow table matching method Match _b , F _up (Match _i ) is the matching method Match _i 's data flow table request amount, n is an integer, i is an integer, and γ _b is the convergence cost of forwarding 1-bit data by the preset switch. The functions of the calculation module symbol and the equalization module basically correspond to the above-mentioned methods, and will not be repeated here.

The preferred embodiment of the present invention has been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the specific details of the above embodiment, within the scope of the technical concept of the present invention, various simple modifications can be made to the technical solution of the present invention, These simple modifications all belong to the protection scope of the present invention.

In addition, it should be noted that the various specific technical features described in the above specific implementation manners may be combined in any suitable manner if there is no contradiction. In order to avoid unnecessary repetition, various possible combinations are not further described in the present invention.

In addition, various combinations of different embodiments of the present invention can also be combined arbitrarily, as long as they do not violate the idea of the present invention, they should also be regarded as the disclosed content of the present invention.

Claims (12)

1. A data flow control method, characterized in that the method comprises: Calculate the data flow table request volume of each switch node in the software-defined network SDN; Optimizing the weight of each link in the SDN according to the aggregation cost of each switch node and the data flow table request amount of each switch node, to balance the data flow of each switch node; The aggregation cost of the nth switch Φ(Match _i )=γ _b Conv(Match _i ), Conv(Match _i )=F _up (Match _b )/F _up (Match _i ), Load(n) is the data forwarding rate of the nth switch, Load _i (n) is the data forwarding rate of the nth switch using the data flow table matching method Match _i , F _up (Match _b ) is the data flow table request amount using the reference data flow table matching method Match _b , F _up (Match _i ) is the request volume of the data flow table using the matching mode Match _i , n is an integer, i is an integer, and γ _b is the aggregation cost of forwarding 1-bit data by the preset switch. 2. method according to claim 1, is characterized in that, calculates Load _i (n) according to following formula: Among them, Match _P is the most accurate data flow table matching method, Match _K is the most relaxed data flow table matching method, P is an integer, K is an integer, K is greater than or equal to P+1, and F _up (n) is the nth The upper limit of the data flow table request amount of the switch, F _up (i) is the flow table request amount within the timeout time T of the data flow table, F _i (n) is the number of data flows forwarded by the nth switch within the timeout time T , F _size is the average size of a single data stream. 3. The method according to claim 2, wherein the data flow table request amount of the nth switch node Where α _b is the ratio of the data flow table request volume of the nth switch using the benchmark matching method Match _b to the total number of data flows forwarded by the switch per unit time. 4. The method according to claim 3, characterized in that, optimizing the weight of each link in the SDN according to the convergence cost of each switch node and the data flow table request amount of each switch node comprises according to each switch node The aggregation cost of the node optimizes the weight of each link in the SDN with the boundary condition that the data flow table request amount of each switch node is less than or equal to the upper limit UB of the data flow table request amount of each switch node. 5. The method according to claim 4, wherein, according to the aggregation cost of each switch node and the data flow table request amount optimization of each switch node, the weight of each link in the SDN comprises optimizing according to the following formula The weight of each link in SDN: $\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\\ \underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; B</span>}\end{array}\right.<span\; class="notranslate">\; .</span>$ 6. The method according to claim 5, characterized in that, according to the aggregation cost of each switch node and the data flow table request amount optimization of each switch node, the weight of each link in the SDN optimization comprises solving the shortest link by genetic algorithm Path to optimize the weight of each link in SDN. 7. A data flow control system, characterized in that the system includes each switch node and a network controller in the software-defined network SDN, and the network controller includes: Calculation module, for calculating the data flow table request amount of each switch node in the software-defined network SDN; A balance module, configured to optimize the weight of each link in the SDN according to the aggregation cost of each switch node and the data flow table request amount of each switch node, so as to balance the data flow of each switch node; The aggregation cost of the nth switch Φ(Match _i )=γ _b Conv(Match _i ), Conv(Match _i )=F _up (Match _b )/F _up (Match _i ), Load(n) is the data forwarding rate of the nth switch, Load _i (n) is the data forwarding rate of the nth switch using the data flow table matching method Match _i , F _up (Match _b ) is the data flow table request amount using the reference data flow table matching method Match _b , F _up (Match _i ) is the request volume of the data flow table using the matching mode Match _i , n is an integer, i is an integer, and γ _b is the aggregation cost of forwarding 1-bit data by the preset switch. 8. system according to claim 7, is characterized in that, computing module calculates Load _i (n) according to following formula: Among them, Match _P is the most accurate data flow table matching method, Match _K is the most relaxed data flow table matching method, P is an integer, K is an integer, K is greater than or equal to P+1, and F _up (n) is the nth The upper limit of the data flow table request amount of the switch, F _up (i) is the flow table request amount within the timeout time T of the data flow table, F _i (n) is the data flow number forwarded by the nth switch within the timeout time T F _size is the average size of a single data stream. . 9. The system according to claim 8, wherein the data flow table request amount of the nth switch node Where α _b is the ratio of the data flow table request volume of the nth switch using the benchmark matching method Match _b to the total number of data flows forwarded by the switch per unit time. 10. The method according to claim 9, wherein the balancing module is further configured to use the aggregation cost of each switch node and the data flow table request amount of each switch node to be less than or equal to the data of each switch node The upper limit of flow table requests UB is the boundary condition to optimize the weight of each link in SDN. 11. The system according to claim 10, wherein the equalization module is also used to optimize the weight of each link in the SDN according to the following formula: $\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; \&Phi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>\\ \underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}{\mathrm{<span\; class="notranslate">\; load</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; match</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; B</span>}\end{array}\right.<span\; class="notranslate">\; .</span>$ 12. The method according to claim 11, wherein the equalization module is also used to optimize the weight of each link in the SDN by solving the shortest path through a genetic algorithm.