CN110784366B - Switch migration method based on IMMAC algorithm in SDN - Google Patents

Switch migration method based on IMMAC algorithm in SDN Download PDF

Info

Publication number
CN110784366B
CN110784366B CN201911097117.5A CN201911097117A CN110784366B CN 110784366 B CN110784366 B CN 110784366B CN 201911097117 A CN201911097117 A CN 201911097117A CN 110784366 B CN110784366 B CN 110784366B
Authority
CN
China
Prior art keywords
controller
switch
load
migration
algorithm
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.)
Active
Application number
CN201911097117.5A
Other languages
Chinese (zh)
Other versions
CN110784366A (en
Inventor
尚凤军
牛新艳
龚汉超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Post and Telecommunications
CERNET Corp
Original Assignee
Chongqing University of Post and Telecommunications
CERNET Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chongqing University of Post and Telecommunications, CERNET Corp filed Critical Chongqing University of Post and Telecommunications
Priority to CN201911097117.5A priority Critical patent/CN110784366B/en
Publication of CN110784366A publication Critical patent/CN110784366A/en
Application granted granted Critical
Publication of CN110784366B publication Critical patent/CN110784366B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08—Configuration management of networks or network elements
    • H04L41/0803—Configuration setting
    • H04L41/0813—Configuration setting characterised by the conditions triggering a change of settings
    • H04L41/082—Configuration setting characterised by the conditions triggering a change of settings the condition being updates or upgrades of network functionality
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08—Configuration management of networks or network elements
    • H04L41/0893—Assignment of logical groups to network elements
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/12—Discovery or management of network topologies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control
    • H04L47/12—Avoiding congestion; Recovering from congestion
    • H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/15—Interconnection of switching modules
    • H04L49/1507—Distribute and route fabrics, e.g. sorting-routing or Batcher-Banyan
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1004—Server selection for load balancing
    • H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/50—Reducing energy consumption in communication networks in wire-line communication networks, e.g. low power modes or reduced link rate

Abstract

The invention belongs to the technical field of software defined networks, and particularly relates to a switch migration method based on an IMMAC algorithm in an SDN (software defined network). the method comprises the steps of obtaining load information of each controller and network topology of the software defined network; calculating the load of a single controller and the average value of the control cluster load according to the load information of each controller; if the average control cluster load value is greater than the control cluster load threshold value and the load of a single controller is less than the load threshold value of the single controller, a switch migration strategy is formulated according to the network topology of the software defined network; optimizing the formulated switch migration strategy by utilizing an IMMAC algorithm, and completing the migration of the switch; the invention can not only reduce the probability that the adjacent controller is easy to overload, but also reduce the time for the high-load controller to respond to the switch request, and compared with the traditional genetic intelligent algorithm, the IMMAC algorithm provided by the invention can improve the time efficiency and the solving precision of the algorithm.

Description

Switch migration method based on IMMAC algorithm in SDN
Technical Field
The invention belongs to the technical field of Software Defined Networking (SDN), and particularly relates to a switch migration method based on an Improved maximum-minimum Ant Colony Algorithm (IMMAC) Algorithm in an SDN.
Background
The internet can be said to be an important foundation for supporting social development and technical progress, and profoundly changes the working, learning and living modes of people, so that people can be in a faster and more convenient environment. The hierarchical structure of the traditional network is the key to the great success of the internet. However, as network coverage continues to increase and network complexity continues to increase, new data is often flooded into the network. According to the data of the Ministry of industry and communications, the access flow of the mobile Internet in China is increased from 12.7 billion GB in 2013 to 711 billion GB in 2018, and the access flow of the mobile Internet of each monthly user is increased from 0.13 GB/month/user in 2013 to 4.42 GB/month/user in 2018. This large internet traffic demand results in network link congestion, reducing network link resource utilization. Moreover, due to the emergence and the rise of technologies such as cloud computing, big data, internet of things and the like, the high coupling and the low cohesion of the traditional network cannot rapidly adapt to new demand changes, and the old architecture greatly hinders the development of new-generation information technologies. Therefore, there is a need for a new network architecture that is capable of dynamically and flexibly managing the network.
In recent years, the appearance of the SDN architecture widens a new field of view of human beings, and provides a new thinking direction for the problem of complex network resource configuration. Different from a traditional network architecture, the main idea of the SDN is to separate a control plane and a forwarding plane, perform centralized control, provide a programmable function, and implement flexible, efficient, and accurate control of link devices. However, new problems to be solved are brought about by the introduction of SDN technology. Among them, reasonable resource scheduling is one of the problems to be solved. Since network resources are limited, it is necessary to fully utilize the limited network resources to provide satisfactory services for human beings in order to meet the increasing demands of people for services. Therefore, how to solve the resource configuration in the SDN network, better utilize the network resources, and ensure the service quality of the network becomes a hotspot and difficulty for people to research.
In the SDN network, the expandability and reliability of the controllers are improved due to the introduction of multiple controllers, and meanwhile, a new problem of unbalanced controller load is brought. Furthermore, there is a static configuration between switches and controllers in an SDN network. Thus, dynamic changes in traffic may result in uneven traffic requests by switches in the SDN network. This deployment is prone to the situation where some controllers are unable to provide sufficient resources to meet the switch requirements, while other controllers are not properly utilizing their resources; this phenomenon can cause network instability, unbalanced controller load, and non-maximized controller resource utilization. Therefore, in order to enable the SDN architecture to implement dynamic resource allocation, it is necessary to propose a new way to solve the problem of controller load imbalance.
Disclosure of Invention
In order to solve the above problem, the present invention provides a switch migration method based on an IMMAC algorithm in an SDN, as shown in fig. 1, including the following steps:
s1, acquiring load information of each controller and network topology of the software defined network;
s2, calculating the load of each controller and the average value of the control cluster load according to the load information of each controller;
s3, if the average value of the load of the control cluster of the controller is smaller than the load threshold of the control cluster and larger than the load threshold of the controller which is smaller than the load threshold of a single controller, maintaining the current situation and not changing;
s4, if the maximum value of the load of the single controller is smaller than the load threshold value of the single controller, the controller is closed or dormant;
s5, if the minimum value of the load of the single controller is larger than the load threshold value of the single controller, adding a new controller;
s6, if the average control cluster load value is larger than the control cluster load threshold value and the load of a single controller is smaller than the load threshold value of the single controller, a switch migration strategy is formulated according to the network topology of the software defined network;
and S7, optimizing the established switch migration strategy by utilizing an IMMAC algorithm, and completing the migration of the switch.
Further, the switch migration policy is expressed as:
min(ω 1 σ+ω 2 cost+ω 3 Υ)
Figure GDA0003596944950000031
Figure GDA0003596944950000032
Figure GDA0003596944950000033
wherein, σ is load balance degree, ω 1 Weight of load balance, cost is migration cost, ω 2 Weight for migration cost, upsilon is the number of switches to be migrated, omega 3 The weight of the number of the switches to be migrated;
Figure GDA0003596944950000034
a load that is a single controller;
Figure GDA0003596944950000035
a load threshold for a single controller; n is the number of controllers in the network; m is the number of switches in the network; t is t ij Representing the values of the elements in the controller-switch mapping relationship matrix.
Further, optimizing the formulated switch migration policy by using an IMMAC algorithm includes:
the IMMAC algorithm corresponds each ant to a solution vector of the optimization problem, namely the optimal mapping relation of the switch and the controller;
inputting a switch set, a controller set and a switch-controller deployment matrix, wherein one row in the switch-controller deployment matrix forms a deployment position of a switch, namely a position point of an ant, and all the rows jointly form an initial path of each ant;
in the iteration of the IMMAC algorithm, each ant further optimizes the initial path according to the change of the concentration of the pheromone on each path and whether the transfer strategy of the switch is met, relieves the overload condition under the initial path, finally selects the optimal path meeting the requirement and outputs the optimal deployment matrix of the switch-controller.
Further, the migration process of the switch includes:
controller C for determining overload A And a switch S requiring migration under the controller x And the idle controller C to which the switch is migrated B ；
Controller C A Sending data packet notification controller C B Starting migration;
controller C B And switch S x Communicate so that the controller C B The role of (1) is changed from Slave controller to Equal controller;
controller C B Sending data packet notification controller C A Preparing for migration;
switch S x Implementation of Slave controller C A Smooth migration to controller C B Up to the controller C A Sending data packet notification controller C B Finishing the migration;
controller C B And switch S x Communicate with each other so that the controller C B The role of (1) is changed from Equal controller to Master controller.
Compared with other modes of directly solving the controller load balance by using a pure switch migration strategy or a single-target optimized intelligent algorithm, the method can reduce the probability that the adjacent controller is easy to overload and reduce the time of the high-load controller for responding to the switch request, and the IMMAC algorithm provided by the invention can greatly improve the algorithm time efficiency and the solving precision compared with the traditional genetic intelligent algorithm.
Drawings
Fig. 1 is a flowchart of a switch migration method based on an IMMAC algorithm in an SDN according to the present invention;
figure 2 is a SDN network topology diagram of a distributed controller of the present invention;
fig. 3 is a process of smooth migration of the switch 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.
The invention provides a switch migration method based on an IMMAC algorithm in an SDN, as shown in figure 1, comprising the following steps:
s1, acquiring load information of each controller and network topology of the software defined network;
s2, calculating the load of each controller and the average value of the control cluster load according to the load information of each controller;
s3, if the average value of the load of the control cluster of the controller is smaller than the load threshold of the control cluster and larger than the load threshold of the controller which is smaller than the load threshold of a single controller, maintaining the current situation and not changing;
s4, if the maximum value of the load of the single controller is smaller than the load threshold value of the single controller, the controller is closed or dormant;
s5, if the minimum value of the load of the single controller is larger than the load threshold value of the single controller, adding a new controller;
s6, if the average control cluster load value is larger than the control cluster load threshold value and the load of a single controller is smaller than the load threshold value of the single controller, a switch migration strategy is formulated according to the network topology of the software defined network;
s7, optimizing the established switch migration strategy by utilizing an IMMAC algorithm, and completing the migration of the switch;
wherein the network topology of the software defined network is shown in figure 2.
In this embodiment, the load information table of the control plane is updated according to the load condition of the controller, and the controller C is confirmed j The process for overloading the controller comprises:
setting SDN network model, wherein M switches S ═ { S } 1 ,S 2 ,...,S M And N controllers C ═ C 1 ,C 2 ,...,C N An SDN network G;
let p i To exchange forMachine S i The number of Packet-In events actually sent to the controller per unit time,
Figure GDA0003596944950000051
for a switch S i The upper limit of the number of Packet-In events that can be sent to the controller per unit time, and therefore the switch S i Can be expressed as
Figure GDA0003596944950000052
Figure GDA0003596944950000053
Is a controller C j The number of packets-In events actually received and to be processed per unit time, i.e., C j The value of the load of (a) is,
Figure GDA0003596944950000054
is a controller C j The upper limit of the number of Packet-In events that can be processed per unit time, namely C j Fuzzy load upper threshold, therefore, controller C j Can be expressed as
Figure GDA0003596944950000055
If controller C is present A According to its fuzzy load threshold
Figure GDA0003596944950000056
Make a judgment if
Figure GDA0003596944950000057
Then confirm controller C A Is an overload controller.
Single controller load, i.e. the number of events of Packet-In actually received and to be processed by the controller per unit time
Figure GDA0003596944950000061
Expressed as:
Figure GDA0003596944950000062
wherein p is i For a switch S i Number of Packet-In events, T, actually sent to the controller In a unit time ij The element of the ith row and the jth column in the matrix of the mapping relation between the switches and the controllers represents the mapping relation between the ith switch and the jth controller;
Figure GDA0003596944950000063
indicates connection to controller C j The number of switches of (c).
In this embodiment, the condition for executing the switch migration operation is determined according to the load condition of the control plane, and the discussion includes four cases:
if it is
Figure GDA0003596944950000064
It is noted that in this case, the controller with the largest load value in the control plane is smaller than its own threshold, and at this time, the load of the entire control plane is too small, and it is not necessary to turn on all the controllers at the same time, so that some controllers can be turned off or dormant to save communication cost and power resources.
If it is
Figure GDA0003596944950000065
It is explained that in this case, although there is an overloaded controller, the actual load value of most controllers is less than its own load threshold. Theoretically, if the SDN network performs frequent switch migration operations, it may cause instability of the entire network, and therefore, it is reasonable to allow a moderate amount of overload, and no switch migration operations are required.
If it is
Figure GDA0003596944950000066
And is
Figure GDA0003596944950000067
Is described hereinIn this case, there may be a plurality of controllers in the control plane in an overloaded state, but not all controllers are overloaded, and in this case, a new controller does not need to be added, and the control plane can reach a balanced state again by performing a switch migration operation.
If it is
Figure GDA0003596944950000068
It is demonstrated that in this case, the controller with the smallest load value in the control plane has exceeded its own threshold, at this time, the entire control plane is heavily overloaded, the switch migration operation has not been able to solve the load imbalance problem, and it is necessary to add a new controller device to the control plane to relieve its load pressure.
The key point of the present invention is that in the third case, the controller load balancing is realized by the switch migration policy, and the making of the switch migration decision includes:
the mapping relationship between the switch and the controller is expressed as T ═ T (T) ij ) M×N 0-1 matrix of (a):
Figure GDA0003596944950000071
wherein a single switch S i Multiple controllers can be connected, but there is only one Master Controller C j To receive and respond to the change machine S i Sending a Packet-In message; is connected to a controller C j Is collected as a switch
Figure GDA0003596944950000072
The number of the switches is
Figure GDA0003596944950000073
Setting the load balance degree of a control plane to describe the load difference condition of controllers in a controller cluster; the smaller the value, the more balanced the control plane load is; conversely, otherwise, it can be measured by mean square error, and the load balancing σ is expressed as:
Figure GDA0003596944950000074
the migration cost at switch migration is cost, which represents the mapping state (T) from the current switch to the controller ij ) M×N Transition to the post-transition New State (T' ij ) M×N The resulting communication overhead, i.e., the average transmission delay. Can be expressed in terms of the average minimum number of transmission hops from all switches to the controller, expressed as:
Figure GDA0003596944950000075
wherein d (C) j ,S j ) For the switch S to be migrated i The minimum number of transmission hops to its Master controller;
Figure GDA0003596944950000076
for the switch S to be migrated i Minimum number of transmission hops to the target Master controller; m is a unit of i Whether the ith switch is changed is represented as:
Figure GDA0003596944950000077
wherein, the exchanger S i From T ═ T ij ) M×N Transition to new post-migration state
Figure GDA0003596944950000078
In the process of (1), if S i If the Master controller is not changed, m i 0, on the contrary, m is changed i ＝1；
The number of switches to be migrated is γ, the less the number of switches to be migrated is, the lower the cost generated by migration is, the smaller the influence on the network is, and the number of switches to be migrated is expressed as:
Figure GDA0003596944950000081
wherein, T ij ' denotes a mapping relationship between the i-th switch and the j-th controller after migration.
In summary, the switch migration policy is expressed as:
min(ω 1 σ+ω 2 cost+ω 3 Υ)
Figure GDA0003596944950000082
Figure GDA0003596944950000083
Figure GDA0003596944950000084
wherein the constraint t ij E {0,1} indicates that the element in the controller-switch can only be 0/1; constraint conditions
Figure GDA0003596944950000085
The Master controller is one and only one for any switch; constraint conditions
Figure GDA0003596944950000086
Indicating that the controller load after migration cannot exceed its upper threshold
Figure GDA0003596944950000087
Namely, the switch is ensured not to cause overload of other controllers after being moved, omega 1 、ω 2 、ω 3 Respectively, the load balance degree, the migration cost during the switch migration, and the weight of the number of switches to be migrated during the switch migration, and ω 1 +ω 2 +ω 3 ＝1。
The invention optimizes the formulated switch migration strategy by utilizing an IMMAC algorithm, wherein the IMMAC algorithm is improved on the basis of a standard ant colony algorithm, and the standard ant colony algorithm comprises the following steps:
(1) initializing relevant parameters including ant colony scale m, pheromone factor alpha, heuristic function factor beta, pheromone volatilization factor rho and pheromone constant tau 0 Maximum number of iterations N max_itera Reading the acquired data into a program, and preprocessing the data;
(2) randomly placing ants at different starting points X i Calculating next visiting city for each ant until all nodes are visited by the ant, wherein the ant individuals start from the starting point X i To the next visiting city X j The random traveling probability of (a) is:
Figure GDA0003596944950000091
wherein the content of the first and second substances,
Figure GDA0003596944950000092
represents that the ant individual k starts from the starting point X when the time is t i To the next visiting city X j Walking transfer feasibility of; allowed k Expressed as the next accessible city node for ant k in the process;
(3) calculating the path length d of each ant k Recording the optimal solution of the current iteration times, and updating the pheromone concentration on the paths at the same time, wherein the change of the pheromone concentration on each path can be determined by the following formula:
τ ij (t)＝(1-ρ)×τ ij (t)+Δτ ij (t)
wherein (1-. rho) is the amount remaining after volatilization of the pheromone,
Figure GDA0003596944950000093
Δτ ij (t) is the amount of pheromone change in the time frame t, representing ant k from starting point X i To destination city X j Pheromone concentration increment between paths;
(4) judging whether the algorithm reaches the maximumLarge number of iterations N max_itera If yes, the algorithm is terminated; otherwise, adding 1 to the iteration times, and returning to the step 2;
(5) and outputting an execution result, and outputting related indexes in the optimizing process according to preset requirements, such as running time, convergence iteration times and the like.
The standard ant colony algorithm is easy to fall into local optimum, and in order to increase the diversity of the colony and obtain a better solution, the concentration of pheromones on each path is limited to [ tau ] min ,τ max ]I.e. the minimum and maximum values that limit the amount of pheromone change, which the present invention improves upon, including:
(1) the ant improves the selection of the next city point in the advancing process, and introduces a roulette mode:
Figure GDA0003596944950000094
Figure GDA0003596944950000095
wherein the roulette wheel probability r 0 E (0,1), r is the probability of random generation at present, and a random factor q is randomly generated after each iteration 0 ∈(0,1)；τ ij (t) pheromone concentration of a link with time t as a starting point and j as an end point; eta ij (t) represents the reciprocal of the distance between links starting at i and ending at j at time t;
Figure GDA0003596944950000101
represents that the ant individual k starts from the starting point X when the time is t i To the next visiting city X j Feasibility of walking transfer.
(2) The value of the volatility factor rho is also one of important factors influencing the size of pheromone quantity on a path, so that the global search capability and the convergence speed of the ant colony algorithm are influenced. In the early stage of the algorithm, the algorithm needs to be converged as soon as possible to find a local optimal solution, and rho needs to be set to be a small value; in the later stage of the algorithm, a large amount of pheromones are accumulated on the path, the concentration of the pheromones is decisive for a calculator of the state transition probability, if the value of rho is too large, the algorithm is easy to fall into a local optimal solution, and in order to search more solutions, the value of rho needs to be set to be a larger value. For selecting an optimal path from the optimal paths by using a greedy strategy, determining the optimal path according to a pheromone concentration change formula of a standard ant colony algorithm, and in order to adapt to a poor link and improve the diversity of a population, the pheromone updating formulas of other paths are expressed as follows:
Figure GDA0003596944950000102
setting a range [ tau ] for pheromone concentration on each path of the ant colony algorithm min ,τ max ]，τ min Increasing the likelihood of searching for a more optimal solution, τ max And the heuristic of experience to the ant colony is ensured.
Applying the improved IMMAC algorithm to solving the SDN switch migration problem comprises the following steps:
first, each ant corresponds to a solution vector of the optimization problem, i.e., the optimal mapping relationship of the switch and the controller. For example: the solution represented by the a-th ant can be expressed as
Figure GDA0003596944950000103
Wherein the content of the first and second substances,
Figure GDA0003596944950000104
indicating a switch S i Is a switch S i With respect to the deployment of N controllers, t ji Indicating a switch S i And a controller C j The deployment relationship between, i.e. t ij 0 represents C j Is S i Equal/Slave Controller, t ij 1 represents C j Is S i Master Controller of (1); m is the number of controllers, N is the number of switches, and the position of each ant is the mapping relation between a group of switches and the controllers. Eyes of a userThe scalar function corresponds to load balance, migration cost and switch migration number, and ω is 1 +ω 2 +ω 3 The magnitude of each weight coefficient can be exchanged according to the importance degree of the objective function;
secondly, in the algorithm, the switch set S ═ S is first input 1 ,S 2 ,...,S i ,...,S M C, controller set C ═ C 1 ,C 2 ,...,C j ,...,C N And the deployment matrix of switch-controller T ═ T (T) ij ) M×N Wherein 0/1 of each row in the deployment matrix T of switch-controllers together form a switch S i I.e. one location point of an ant, all rows together make up the initial path of each ant. In the iteration process, each ant further optimizes the initial path according to the change of the concentration of the pheromone on each path and whether the target function is met, relieves the overload condition under the initial path, and finally selects the optimal path meeting the requirement;
and finally, outputting an optimal 0-1 mapping relation matrix between the controllers and the switches.
As shown in fig. 3, the migration process of the switch includes:
controller C for determining overload A And a switch S requiring migration under the controller x And the idle controller C to which the switch is migrated B ；
Controller C A Sending data packet notification controller C B Starting migration;
controller C B And switch S x Communicate so that the controller C B The role of (1) is changed from Slave controller to Equal controller;
controller C B Sending data packet notification controller C A Preparing for migration;
switch S x Implementation of Slave controller C A Smooth migration to controller C B Up to the controller C B Sending data packet notification controller C A Finishing the migration;
controller C B And exchange S x To communicate with each other and to communicate with each other,so that the controller C B The role of (1) is changed from Equal controller to Master controller.
This example simulates an SDN network of 5 controllers and 20 switches. And one controller is overloaded by increasing the rate at which part of the switches send Packet-In, resulting In the data shown In table 1.
Table 1 flow arrival rate and response time before performing a migration policy
Figure GDA0003596944950000121
Once the arrival rate of Packet-In of a certain switch exceeds the capacity of the controller, the response time thereof increases sharply. Suppose that
Figure GDA0003596944950000122
And is
Figure GDA0003596944950000123
At this point, the control plane may have multiple controllers in an overloaded state, but not all controllers are overloaded, and at this point, no new controllers need to be added, and the control plane can reach a balanced state again by performing switch migration operations.
The technology mainly solves the problem of unbalanced load of the controller under the condition. Therefore, for the switch migration policy, the load balance degree has the greatest influence on the switch migration policy, that is, the weight coefficient is set relatively highest, so the weight coefficients can be respectively set to ω 1 ＝0.5，ω 2 ＝0.25，ω 3 The data after the inventive strategy was performed is as in table 2, 0.25.
Table 2 flow arrival rate and response time after execution of migration policy
Figure GDA0003596944950000124
It can be seen In table 1 that the controller continues to increase the response time of the controller when the Packet-In Packet arrival rate is 50, but it is found through table 2 that the controller performs the switch migration operation when the Packet-In Packet arrival rate is 50, so that the response time is reduced and gradually levels off.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (8)

  1. The method for migrating the switch based on the IMMAC algorithm in the SDN is characterized by comprising the following steps of:
    s1, acquiring load information of each controller and network topology of the software defined network;
    s2, calculating the load of each controller and the average value of the control cluster load according to the load information of each controller;
    s3, if the average value of the load of the control cluster of the controller is smaller than the load threshold of the control cluster and the load of the controller is larger than the load threshold of a single controller, maintaining the current situation and not changing;
    s4, if the maximum value of the load of the single controller is smaller than the load threshold value of the single controller, the controller is closed or dormant;
    s5, if the minimum value of the load of the single controller is larger than the load threshold value of the single controller, adding a new controller;
    s6, if the average control cluster load value is larger than the control cluster load threshold value and the load of a single controller is smaller than the load threshold value of the single controller, a switch migration strategy is formulated according to the network topology of the software defined network;
    s7, optimizing the established switch migration policy by using an IMMAC algorithm, i.e., an improved max-min ant colony algorithm, and completing the migration of the switch, specifically including:
    the IMMAC algorithm corresponds each ant to a solution vector of the optimization problem, namely the optimal mapping relation of the switch and the controller;
    inputting a switch set, a controller set and a switch-controller deployment matrix, wherein one row in the switch-controller deployment matrix forms a deployment position of a switch, namely a position point of an ant, and all the rows jointly form an initial path of each ant;
    in the iteration of the IMMAC algorithm, each ant further optimizes the initial path according to the change of the concentration of the pheromones on each path and whether the transfer strategy of the switch is met, relieves the overload condition under the initial path, finally selects the optimal path meeting the requirement and outputs the optimal deployment matrix of the switch-controller;
    the IMMAC algorithm is improved on the basis of a standard ant colony algorithm, and the improvement specifically comprises the following steps:
    during the course of the ants' travel, the introduction of roulette improved the process of ants selecting the next point, expressed as:
    Figure FDA0003727636840000021
    Figure FDA0003727636840000022
    the concentration of pheromones on each path is set to a range [ tau ] min ,τ max ]After each iteration, the optimal path is selected by using a greedy strategy to carry out pheromone updating, and the pheromone updating is represented as follows:
    τ ij (t+1)＝(1-ρ)×τ ij (t)+Δτ ij (t)；
    the pheromone updates on the other paths are represented as:
    Figure FDA0003727636840000023
    wherein, X j Representing nodes visited by ants; alpha is an pheromone factor; beta is a heuristic function factor; rho is a volatilization factor, and m is the ant colony scale; tau is ij (t +1) representsPheromones for the t +1 th iteration;
    Figure FDA0003727636840000024
    represents that the ant individual k starts from the starting point X when the time is t i To the next visiting city X j Walking transfer feasibility of; allowed k Expressed as the next accessible city node for ant k in the process; delta tau ij (t) is the amount of change in pheromone over time t, expressed as
    Figure FDA0003727636840000025
    Δτ ij k (t) represents the departure point X of ant k i To destination city X j Inter-pathway pheromone concentration increments; n is a radical of itera Representing the current number of iterations, N Max_itera Maximum number of iterations; eta ij Represents the inverse of the distance between two points on link ij at time t; r is the random probability of the roulette wheel; r is 0 Is the roulette wheel probability; q. q.s 0 Is a randomly generated random factor after each iteration.
  2. 2. The IMMAC algorithm-based switch migration method In SDN, according to claim 1, wherein the load of a single controller is the number of Packet-In events actually received and required to be processed by the controller per unit time
    Figure FDA0003727636840000026
    Expressed as:
    Figure FDA0003727636840000031
    wherein p is i For a switch S i Number of Packet-In events, T, actually sent to the controller In a unit time ij The element of the ith row and the jth column in the matrix of the mapping relation between the switches and the controllers represents the mapping relation between the ith switch and the jth controller;
    Figure FDA0003727636840000032
    indicates connection to controller C j The number of switches of (c).
  3. 3. The method for migrating switches in an SDN based on an IMMAC algorithm according to claim 2, wherein the matrix T of the mapping relationship between the switches and the controllers is represented as: t ═ T (T) ij ) M×N Mapping relation T between ith switch and jth controller ij Expressed as:
    Figure FDA0003727636840000033
    wherein, C j Represents the jth controller; s i Represents the ith switch; n is the number of controllers in the network; m is the number of switches in the network.
  4. 4. The method of claim 1, wherein the switch migration policy is expressed as:
    min(ω 1 σ+ω 2 cost+ω 3 Υ)
    Figure FDA0003727636840000034
    Figure FDA0003727636840000035
    Figure FDA0003727636840000036
    wherein, σ is load balance degree, ω 1 Weight of load balance, cost is migration cost, ω 2 Weight for migration cost, upsilon is the number of switches to be migrated, omega 3 The weight of the number of the switches to be migrated;
    Figure FDA0003727636840000037
    a load that is a single controller;
    Figure FDA0003727636840000038
    a load threshold for a single controller; n is the number of controllers in the network; m is the number of switches in the network; t is ij Indicating a switch S i And a controller C j The mapping relationship between them.
  5. 5. The method of claim 4, wherein the load balancing σ is expressed as:
    Figure FDA0003727636840000041
    wherein the content of the first and second substances,
    Figure FDA0003727636840000042
    representing the number of events of Packet-In actually received and needed to be processed by the controller per unit time.
  6. 6. The IMMAC algorithm-based switch migration method in an SDN according to claim 4, wherein the migration cost is expressed as:
    Figure FDA0003727636840000043
    wherein d (C) j ,S j ) For the switch S to be migrated i Minimum number of transmission hops to its Master controller;
    Figure FDA0003727636840000044
    for the switch S to be migrated i Minimum number of transmission hops to the target Master controller; m is i Indicating whether the ith switch is changed;
    Figure FDA0003727636840000045
    is a controller C j The number of switches connected.
  7. 7. The method of claim 4, wherein the number y of switches to be migrated is represented as:
    Figure FDA0003727636840000046
    wherein, T ij ' denotes a mapping relationship between the i-th switch and the j-th controller after migration.
  8. 8. The method of claim 1, wherein the switch migration process based on IMMAC algorithm comprises:
    controller C for determining overload A And a switch S requiring migration under the controller x And the idle controller C to which the switch is migrated B ；
    Controller C A Sending data packet notification controller C B Starting migration;
    controller C B And switch S x Communicate so that the controller C B The role of (1) is changed from Slave controller to Equal controller;
    controller C B Sending data packet notification controller C A Preparing for migration;
    switch S x Implementation of Slave controller C A Smooth migration to controller C B Up to the controller C A Sending data packet notification controller C B Finishing the migration;
    controller C B And switch S x Communicate with each other so that the controller C B The role of (1) is changed from Equal controller to Master controller.
CN201911097117.5A 2019-11-11 2019-11-11 Switch migration method based on IMMAC algorithm in SDN Active CN110784366B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911097117.5A CN110784366B (en) 2019-11-11 2019-11-11 Switch migration method based on IMMAC algorithm in SDN

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911097117.5A CN110784366B (en) 2019-11-11 2019-11-11 Switch migration method based on IMMAC algorithm in SDN

Publications (2)

Publication Number Publication Date
CN110784366A CN110784366A (en) 2020-02-11
CN110784366B true CN110784366B (en) 2022-08-16

Family

ID=69391761

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911097117.5A Active CN110784366B (en) 2019-11-11 2019-11-11 Switch migration method based on IMMAC algorithm in SDN

Country Status (1)

Country Link
CN (1) CN110784366B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111565214A (en) * 2020-03-06 2020-08-21 国网重庆市电力公司南岸供电分公司 Software defined network load balancing method, system and device
CN111711576B (en) * 2020-06-30 2022-03-04 西安电子科技大学 Controller load balancing system and method based on efficient switch migration
CN113328889B (en) * 2021-05-31 2022-06-28 河南财政金融学院 Distributed optimization method for control layer architecture in software defined network
CN113504976B (en) * 2021-06-10 2023-05-23 中国联合网络通信集团有限公司 Scheduling method, system, terminal equipment and storage medium for software-defined network architecture
CN114928614B (en) * 2022-05-16 2023-05-23 济南大学 Deterministic network load balancing method and system based on SDN
CN115499376B (en) * 2022-07-29 2024-01-02 天翼云科技有限公司 Load balancing method, system, electronic equipment and storage medium
CN115361341B (en) * 2022-10-19 2023-03-24 南京邮电大学 SDN multi-controller-based data center network load balancing method

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109933425A (en) * 2019-01-31 2019-06-25 南京邮电大学 A kind of cloud computing virtual machine placement method based on improvement ant group algorithm
CN110298589A (en) * 2019-07-01 2019-10-01 河海大学常州校区 Based on heredity-ant colony blending algorithm dynamic Service resource regulating method

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104317646B (en) * 2014-10-23 2017-10-24 西安电子科技大学 Based on cloud data center dispatching method of virtual machine under OpenFlow frameworks
CN104700251B (en) * 2015-03-16 2018-02-02 华南师范大学 The improvement minimax ant colony optimization method and system of a kind of vehicle dispatching problem
US9756121B2 (en) * 2015-06-24 2017-09-05 International Business Machines Corporation Optimizing routing and load balancing in an SDN-enabled cloud during enterprise data center migration
CN107276662B (en) * 2017-07-27 2019-12-03 大连大学 A kind of software definition Information Network multi-controller dynamic deployment method
CN108900428A (en) * 2018-06-26 2018-11-27 南京邮电大学 Controller load-balancing method based on interchanger dynamic migration

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109933425A (en) * 2019-01-31 2019-06-25 南京邮电大学 A kind of cloud computing virtual machine placement method based on improvement ant group algorithm
CN110298589A (en) * 2019-07-01 2019-10-01 河海大学常州校区 Based on heredity-ant colony blending algorithm dynamic Service resource regulating method

Also Published As

Publication number Publication date
CN110784366A (en) 2020-02-11

Similar Documents

Publication Publication Date Title
CN110784366B (en) Switch migration method based on IMMAC algorithm in SDN
CN112134916B (en) Cloud edge collaborative computing migration method based on deep reinforcement learning
CN110275758B (en) Intelligent migration method for virtual network function
CN111953758B (en) Edge network computing unloading and task migration method and device
CN111130904B (en) Virtual network function migration optimization algorithm based on deep certainty strategy gradient
CN111694636B (en) Electric power Internet of things container migration method oriented to edge network load balancing
CN111538587B (en) Service function chain reconfiguration method based on load balancing
WO2023040022A1 (en) Computing and network collaboration-based distributed computation offloading method in random network
CN111988225B (en) Multi-path routing method based on reinforcement learning and transfer learning
CN111984419B (en) Complex task computing migration method for edge environment reliability constraint
CN111010295B (en) SDN-MEC-based power distribution and utilization communication network task migration method
CN109104464A (en) A kind of distributed data update method towards collaboration storage under edge calculations environment
CN109167671A (en) A kind of adapted communication system equally loaded dispatching algorithm towards quantum key distribution business
CN114567598A (en) Load balancing method and device based on deep learning and cross-domain cooperation
CN111988787B (en) Task network access and service placement position selection method and system
Kumar et al. Using clustering approaches for response time aware job scheduling model for internet of things (IoT)
CN111885493B (en) Micro-cloud deployment method based on improved cuckoo search algorithm
CN114024970A (en) Power internet of things work load distribution method based on edge calculation
CN113965569B (en) High-energy-efficiency low-delay edge node calculation migration configuration system
CN111510319A (en) Network slice resource management method based on state perception
Liu et al. Network function migration in softwarization based networks with mobile edge computing
Baek et al. FLoadNet: Load Balancing in Fog Networks With Cooperative Multiagent Using Actor–Critic Method
CN115361453A (en) Load fair unloading and transferring method for edge service network
CN115118728A (en) Ant colony algorithm-based edge load balancing task scheduling method
CN114785692A (en) Virtual power plant aggregation regulation and control communication network flow balancing method and device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant