CN108777636A - A kind of multi-controller Optimization deployment method of robust in software defined network - Google Patents
A kind of multi-controller Optimization deployment method of robust in software defined network Download PDFInfo
- Publication number
- CN108777636A CN108777636A CN201810518568.0A CN201810518568A CN108777636A CN 108777636 A CN108777636 A CN 108777636A CN 201810518568 A CN201810518568 A CN 201810518568A CN 108777636 A CN108777636 A CN 108777636A
- Authority
- CN
- China
- Prior art keywords
- controller
- node
- network
- rcm
- algorithms
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- 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
Abstract
A kind of multi-controller Optimization deployment method of robust in software defined network, including：S100：Analyze network topology structure；S200：Build Related Mathematical Models；S300：The selection and solution of approximate data.SDN network multi-controller deployment issue is abstracted and turns to a graph theoretic problem by this method, converts the graph theoretic problem to integral linear programming problem by founding mathematical models, the strategy solved to the problem using approximate data.The method makes SDN network figure not only meet given controller coverage rate under the situation for arbitrarily removing k side, but also can take into account network transmission efficiency and load balancing.
Description
Technical field
The disclosure belongs to network technique field, more particularly to the multi-controller Optimization Dept. of robust in a kind of software defined network
Arranging method.
Background technology
Software defined network (Software defined network, SDN) uses control plane and data plane phase
The network architecture of separation realizes the flexible control of network data forwarding.Different from traditional network, control plane is as SDN nets
The Optimization deployment of the core of network, controller involves the various aspects such as network delay, load and internet security；With traditional network class
Seemingly, SDN network is equally faced with the network robustness problem of network node or link failure under accident.
Since controller has undertaken the control work of whole network, the processing capacity and controller and interchanger of controller it
Between the time delay that communicates have important influence to the performance of whole network.For catenet, flowed by single controller
The distribution of table can not be competent at the demand of all interchangers, it is necessary to share whole system using distributed multiple controllers
Pressure, however how to dispose multiple controllers in a network, i.e. controller deployment (Control Placement, CP) is still
One open problem；The entirely controlled plane control processed of SDN network forwarding unit, easily causes to control under network failure situation
Communication disruption between plane and Forwarding plane, and then influence the normal operation of SDN network；Especially in SDN network node or chain
Under the accident of road failure, the especially needed concern of robustness of controller deployment.Therefore, research reply event promotes SDN nets
The controller optimization deployment issue of network robustness is of great significance.
In the deployment of practical multi-controller, in order to reduce lower deployment cost, controller number should be reduced to the greatest extent.Ensure it is low at
This while, will also improve network performance as far as possible, the average delay of interchanger to controller answer it is as small as possible, between each controller
Load should be as balanced as possible, thus existing controller deployment strategy considers mostly that controller number is few, time delay is small, load balancing
Target, rarely have consider solution of emergent event controller deployment robustness problem.SDN network meets with failure or attack
In the case of, it is highly that how to be configured by limited controller optimization, which improves the biological treatability of SDN network controller service,
Research.
Invention content
To solve the above-mentioned problems, present disclose provides a kind of multi-controller Optimization deployments of robust in software defined network
Method includes the following steps：
S100：Analyze network topology structure
S101：Network topology structure is analyzed, the multi-controller deployment issue in software defined network SDN is abstracted and turns to one
A graph theoretic problem based on undirected graph；
S102：It proposes a kind of new measurement, coverage rate index is controlled, for indicating that the interchanger that controller can be accessed is total
Number accounts for the ratio of all interchanger sums in network；
S200：Build Related Mathematical Models
A undirected graph G (V, L) and controller coverage rate Cpr are given, is placed by reasonable deployment controller
Set of node C so that the undirected graph is labeled as under the situation for arbitrarily removing k sideBoth full
Given controller coverage rate Cpr enough, and network transmission efficiency and load balancing can be taken into account, wherein V indicates interchanger set,
Indicated with node in undirected graph, and n indicates the number of node, i.e. n=| V |, L indicates the link set of connection interchanger
It closes, is indicated with side in undirected graph, C expression controller node set, i.e., the telephone net node set that controller is connected,
The scene set that any limit is removed in undirected graph G, is denoted as S, removes the situation set on k side, is denoted as sk,
The worst scene for removing k side is denoted as sb, sb ∈ sk,Indicate the link set of the arbitrary connection interchanger for removing k side,Indicate to remove the link set of the connection interchanger on k side under the worst situation, k be positive integer and 1≤k≤| L |；
S300：The selection and solution of approximate data
The k-RCM near-optimization algorithms for taking into account network delay and the robust of load balancing are chosen, k-RCM algorithms include mainly
The worst scene moves down the GN algorithms of flash trimming, is denoted as k-RCM-GN, and Dual Approximate is denoted as k-RCM-DOLP and remote exchange
Machine Node extraction algorithm, is denoted as k-RCM-outliers.
Through the above technical solutions, the disclosure can rationally dispose optimal controller resource so that SDN network is arbitrary
It is as small as possible to the average delay between the average delay and controller of controller to remove interchanger under the situation of k link,
And keeping the load of each controller distribution as balanced as possible, the controller for improving robustness places problem.
Description of the drawings
Fig. 1 is the multi-controller Optimization Dept. of robust in a kind of software defined network provided in an embodiment of the present disclosure
The flow diagram of arranging method；
Fig. 2 is the flow diagram of k-RCM near-optimizations algorithm in an embodiment of the present disclosure；
Fig. 3 is the flow diagram of k-RCM-DOLP algorithms in an embodiment of the present disclosure；
Fig. 4 a-4b are that OS3E network controllers place comparative result figure in an embodiment of the present disclosure；
Fig. 5 is the loading condition of OS3E network branches key-machine in an embodiment of the present disclosure；
Fig. 6 is the loading condition of OS3E networks k-RCM-DOLP algorithms in an embodiment of the present disclosure；
Fig. 7 be in an embodiment of the present disclosure OS3E networks successively cut edge when control overlay node number comparison diagram；
Fig. 8 is OS3E networks stochastic simulation cut edge probability comparison diagram in an embodiment of the present disclosure；
Fig. 9 is the TE index comparison diagrams in OS3E networks under the worst scene for removing side in an embodiment of the present disclosure；
Figure 10 is the configuration of US Carrier networks k-RCM approximate data controllers and load knot in an embodiment of the present disclosure
Fruit；
Figure 11 is that US Carrier networks remove side control overlay node number comparison diagram successively in an embodiment of the present disclosure；
Figure 12 is TE index pair of the US Carrier networks under the worst scene for removing side in an embodiment of the present disclosure
Than figure.
Specific implementation mode
To solve the multi-controller deployment issue of robust in SDN network, the robust control for initially setting up k- link failures is needed
The integer linear rule of device deployment (Robust Control Placement for k-links-failure, k-RCM) problem processed
(Integer linear programming, ILP) model is drawn, which provides by distributing the controller of minimum cost rationally
Source, to meet the control coverage rate of controller in SDN network in the case of arbitrary k- link failures.It is solved to further facilitate, it is right
The integral linear programming model relaxes, and establishes the linear programming model of k-RCM, then utilizes the antithesis of linear programming problem
Planning, and uses for reference facility addressing Solve problems thinking, using the robust for taking into account network delay and load balancing k-RCM approximations most
Excellent algorithm, referred to as k-RCM algorithms, k-RCM algorithms can significantly reduce the computational complexity of direct solution k-RCM problem models.
Referring to Fig. 1, in one embodiment, it discloses a kind of multi-controller Optimization Dept.s of robust in software defined network
Arranging method includes the following steps：
S100：Analyze network topology structure
S101：Network topology structure is analyzed, the multi-controller deployment issue in software defined network SDN is abstracted and turns to one
A graph theoretic problem based on undirected graph；
S102：It proposes a kind of new measurement, coverage rate index is controlled, for indicating that the interchanger that controller can be accessed is total
Number accounts for the ratio of all interchanger sums in network；
S200：Build Related Mathematical Models
A undirected graph G (V, L) and controller coverage rate Cpr are given, is placed by reasonable deployment controller
Set of node C so that the undirected graph is labeled as under the situation for arbitrarily removing k sideBoth full
Given controller coverage rate Cpr enough, and network transmission efficiency and load balancing can be taken into account, wherein V indicates interchanger set,
It is indicated with node in undirected graph, and n indicates that the number of node, i.e. n=V, L indicate the link set of connection interchanger,
It is indicated with side in undirected graph, C expression controller node set, i.e., the telephone net node set that controller is connected,
The scene set that any limit is removed in undirected graph G, is denoted as S, removes the situation set on k side, is denoted as sk,It moves
Except the worst scene on k side is denoted as sb, sb∈sk,Indicate the link set of the arbitrary connection interchanger for removing k side,
Indicate to remove the link set of the connection interchanger on k side under the worst situation, k be positive integer and 1≤k≤| L |；
S300：The selection and solution of approximate data
The k-RCM near-optimization algorithms for taking into account network delay and the robust of load balancing are chosen, k-RCM algorithms include mainly
The worst scene moves down the GN algorithms of flash trimming, is denoted as k-RCM-GN, and Dual Approximate is denoted as k-RCM-DOLP and remote exchange
Machine Node extraction algorithm, is denoted as k-RCM-outliers.
In another embodiment, k-RCM problems are modeled as integral linear programming (integer linear
Programming, ILP) model, referred to as k-RCM_ILP models.It includes the following contents to model k-RCM_ILP models：
Object function：
Constraints：
Wherein：
n：Indicate the quantity of SDN network interior joint；
Sk：Indicate all situation set on k side of removal；
sk：Indicate the situation set on k side of removal,
sb：The worst scene on k side of the corresponding removal；
Scene skThe probability of appearance,
fj：The cost of controller is placed on node j；
In scene skUnder, general switch node i is connected to the time delay that controller places node j；
Cpr：Indicate given SDN controller coverage rates；
xj：It is a 0-1 variable, xj=1 indicates that node j places controller by selection, is otherwise 0；
It is a 0-1 variable, indicates node i in scene skIn whether node j covering placed by controller, if
It is, thenOtherwise
It is a 0-1 variable, indicates node i in scene skIn whether be remote telephone net node, if it is,Otherwise
I, j indicate any one interchanger, i.e. i in interchanger set, j ∈ V, k be positive integer and 1≤k≤| L |.
It should be noted that above-mentioned integral linear programming simulated target is by the minimum controller node of Optimization deployment,
So that SDN network also can guarantee the requirement of control coverage rate Cpr under the arbitrary situation for removing k side, and take into account improvement SDN
Network-based control device efficiency of transmission.In target function type (L1-1), the time delay index of node is usedTo control candidate configuration section
The efficiency of transmission of point；Use controller configuration cost fjWhile parameter so that controlling coverage rate is not less than Cpr, make controller
Configure Least-cost.
Constraint formula (L1-2) controls in all situations, could be by other after node j has been selected as controller placement node
General switch node i is connected on the controller.If xj=0, theni∈V；If xj=1, thenOr 0i
∈V。
Constraint formula (L1-3) ensures in all situations, otherwise node is controlled telephone net node or is remote
Telephone net node.It illustrates, when node j, which is selected as controller, places node
Constraint formula (L1-4) is very important constraints, it is indicated that the SDN network control in the case where removing k side situation is covered
Lid rate should be not less than given SDN network control coverage rate Cpr.
Finally, constraint formula (L1-5) shows variable xj、WithAll it is 0-1 variables.
Controller is placed due to finding part of nodes in SDN network so that controller is in the arbitrary scene for breaking k side
Meet the controller coverage rate requirement of Cpr, which is proved to belong to NPC problems.For simplify problem solving, model modification be
Under the worst scene for removing k side, limited controller resource how is distributed rationally so that SDN network can meet given
Coverage rate demand is controlled, and network transmission efficiency and load balancing can be taken into account.
In another embodiment, it is assumed that the worst cut edge scene sbProbability of happeningRemoving k side most
Under bad scene, the integral linear programming model k-RCM_ILP of k-RCM problems is as follows：
Object function：
Constraints：
Wherein, i, j indicate any one interchanger, i.e. i in interchanger set, j ∈ V.
In another embodiment, to solve L2, constraint relaxation is carried out to 0-1 integers, establishes its corresponding linear programming
Model k-RCM_LP is：
Object function：
Constraints：
Wherein, i, j indicate any one interchanger, i.e. i in interchanger set, j ∈ V.
Preferably, it is contemplated that integer solving complexity is relatively high, and Integer constrained characteristic, which is revised as linear restriction, can obviously reduce
Algorithm complexity, so by constraints
It replaces with
In another embodiment, using the dual program of linear programming problem, its corresponding dual program model is established
K-RCM-DOLP is：
Object function：
max∑i∈Vαi+n(α-1)q (L4-1)
Constraints：
∑i∈Vβij≤fjj∈V (L4-2)
Wherein, dual variable βijIndicate that the controller node j that telephone net node i connects it places cost fjContribution
Value；Dual variable αiIndicate the wastage in bulk or weight cost of telephone net node i, wastage in bulk or weight cost includes that interchanger i connects its controller j
Time delay cijCost f is placed with the controller that interchanger i connects itjContribution margin βij；Dual variable q indicates interchanger wastage in bulk or weight
Cost αiMaximum value；I, j indicate any one interchanger, i.e. i in interchanger set, j ∈ V；N indicates of nodes
Number.
Controller is a kind of server resource, needs to be connected on certain interchanger, so controller resource optimization is disposed
Position refer to the telephone net node that controller is connected position.In addition, in order to further decrease controller resource deployment
Cost optimizes the efficiency of network-control management, introduces control coverage rate index Cpr, which can be accessed controller
Interchanger sum account for the ratios of all interchanger sums in network.Cpr index definitions are Cpr=nc/ n, wherein ncIndicate energy
The control overlay node sum of controller resource is accessed, i.e., it includes the general switch node of energy access controller resource,
Also include the controller node itself for deploying controller resource, n indicates all interchanger sums in network.Given meet demand
Control coverage rate show not requiring the interchanger in network to be completely covered by controller under accident, do not influencing network base
In the case of this service function, extremely a other remote interchanger is allowed (also referred to as " outlies ") to access less than controller.
Betweenness center the index reflection information processing capability and message data rate of node and side.Side betweenness center
Arbitrary node is defined as in network to the number based on shortest path by all paths on the side.K side is removed in this method
The worst situation in each remove side, be all to put in order one by one according to the descending of betweenness center index in side in network
What removal obtained.
It is in another embodiment, specific as follows it discloses k-RCM near-optimization algorithms referring to Fig. 2：
S301：Given undirected graph G (V, L) and controller coverage rate Cpr；
S302：Assuming that the maximum value that the controller in the known optimal deployment scheme of controller places cost is f ', f is changedj：
Work as fjWhen > f ', f is enabledj=∞, otherwise fjIt is constant, wherein fjIndicate the cost of placement controller on node j；
S303：Based on k-RCM-DOLP algorithms, remote node is extracted, remote set of node is deleted in G；
S304：Based on k-RCM-GN algorithms, the worst disconnected k side situation is found；
S305：Based on k-RCM-DOLP algorithms, the load table that controller places node set and each controller is generated；
S306：It is arranged to make remote node to be controlled by the controller of remote node and is connected thereto the control of expense minimum
Device.
It in another embodiment, can be with most while extracting community structure in view of classical community discovery GN algorithms
Network is divided into relatively uniform subnet by fast speed, and the k-RCM-GN algorithms proposed based on GN algorithms may be implemented in SDN nets
The worst situation for removing k side in network is portrayed, and then k-RCM algorithms is supported to improve controller portion while taking into account load balancing
Robustness of the node to link fails is affixed one's name to, it is specific as follows：
A) initial setting up, num=0,
B) in given undirected graph it is all while by while betweenness center be ranked up, then deleted in undirected graph
The maximum side l of flash trimming betweenness center index；
C) num=num+1 is corrected,
D) check whether num is equal to k, if num<K, then repeatedly step b and step c；If num=k is thened follow the steps
e；
E) collection on the k side that the worst scene of SDN network interrupts is combined into
When there is community structure in SDN network, k-RCM-GN algorithms can extract community structure.Utilize community structure
Density further selects the controller of efficiency of transmission highest and Least-cost to match using k-RCM-DOLP algorithms in corporations
Set node set.If the arbitrary node in community structure is controlled by the controller node in the corporations, can significantly carry
High controller node faces robustness when link fails.When in SDN network without apparent community structure, k-RCM-GN algorithms
Network is divided into most fast speed and more uniform multiple is not connected to subnet.Further utilize k-RCM-DOLP algorithms each
It selects efficiency of transmission optimal in subnet and the controller deployment scheme of Least-cost, controller node management in each subnet is made to correspond to
Telephone net node in subnet, the problem of to can effectively improve controller load balancing.
In another embodiment, telephone net node remote in SDN network (outliers) disposes controller optimization
Large effect is will produce, the efficiency of transmission of network can be not only reduced, also improvement network robustness can be made to pay a high price.
When removing k side under the worst scene, to make controller cover node as much as possible, using k-RCM-outliers algorithms.
Controller is placed in the case where not considering that remote node influences, to reduce influence of the remote node to controller coverage rate.
The k-RCM-outliers algorithms are specific as follows：
A) assume that the maximum value that the controller in the optimal deployment scheme of known controller places cost is f ', change fj：Work as fj
When > f ', f is enabledj=∞, otherwise fjIt is constant, wherein fjIndicate the cost of placement controller on node j；
B) k-RCM-DOLP algorithms are run, are less than (1-Cpr) * when having in network | V | a node is not determined as controlling
When device or the network equipment controlled, then k-RCM-DOLP algorithms are terminated, remaining node is then the remote section being selected
Point；
C) these remote nodes are connected on the controller of nearest not load saturation.
Remote node is connected on the controller of nearest not load saturation by above-mentioned algorithm, to realize to remote section
The management and control of point.
Referring to Fig. 3, in another embodiment, to find minimum optimal controller configuration set, meet given SDN
Network-control coverage rate and guarantee network transmission efficiency, use Dual Approximate k-RCM-DOLP.The k-RCM-DOLP
Algorithm is specific as follows：
a)αi=0, βij=0, mark all αiIt can increase, βijIt can not increase；
B) to increasable αiThere is αi=αi+1；
C) meet α with the presence or absence of (i, j)i=cij；And if so, whether decision node j, which has been selected as, is placed controller
Node, if it is flag node i controlled by node j；If otherwise label (i, j) connection side is tight, and βijIt is followed in next time
It can increase in ring；If there is no then executing next step；
D) to increasable βijThere is βij=βij+1；
E) meet restrictive condition ∑ with the presence or absence of node ji∈Vβij=fj(j∈V)；And if so, from above-mentioned limit is met
Device places node to the node j ' for selecting the attachable network equipment most in the node of condition processed in order to control；If there is no then holding
Row step g；
F) label can be by the network equipment of j ' controls, and βijAnd αiDo not increasing, i ∈ { can be set by the network of j ' controls
It is standby }；
G) whether all nodes have been selected, if then terminating this algorithm, step b is executed if otherwise redirecting.
Following example combination attached drawing 4-12 is illustrated.
In one embodiment, in the OS3E networks as shown in Fig. 4 a-4b, it is assumed that on the highest node of node efficiency 13
It placed a controller, under proper network environment, the control on node 13 can be efficiently accessed in all nodes in network
Device processed.It is arranged based on side betweenness center index descending, is removing 9 sides (i.e.
)
The worst scene under, be based on k-RCM algorithms, k-median approximate datas and branch-bound algorithm Optimization deployment OS3E network-controls
The capability analysis of device is as follows.
Assuming that given control coverage rate is 85%, the optimal content configuration node collection that branch-bound algorithm obtains is combined into C
={ 2,13,16,28,31 }, loading condition are shown in attached drawing 5；Under same case, the optimal controller obtained based on k-RCM algorithms is matched
It is C={ 2,6,13,29,32 } to set node set, and loading condition is shown in attached drawing 6.In Fig. 4 a × indicate be that k-median algorithms are examined
The controller distribution situation being delayed when considering the worst, zero expression is that k-median algorithms consider that controller when average delay is distributed
Situation；In Fig. 4 b × and indicate the obtained controller distribution situation of branch-bound algorithm, zero indicate is control that k-RCM algorithms obtain
Device distribution situation processed.
Robustness：
Fig. 7 describe control coverage rate index with the worst scene moves down the increase of flash trimming number and downward trend, lead to
It crosses and compares k-median algorithms based on average delay, k-median algorithms based on the worst time delay, branch-bound algorithm and based on k-
The variation of the control coverage rate index of the controller configuring condition of RCM algorithm configurations, it was found from data analysis in figure：Based on k-
The controller configuration result of RCM algorithms can significantly improve controller coverage rate of the OS3E networks in k link failures.
Fig. 8 describe according to based on k-median algorithms average delay, it is fixed based on the worst time delay of k-median algorithms, branch
Boundary's algorithm and controller configuring condition based on k-RCM algorithm configurations, in OS3E networks, simulation 105Secondary disconnected l=at random 1,
2,3 ... 10 } experiment on side is counted 105In secondary experiment, control coverage rate requires horizontal probability value less than Cpr.From figure
Known to data analysis：In stochastic simulation experiment comparison k-median algorithms consider average delay algorithm, branch-bound algorithm and
K-RCM algorithm ratio k-median algorithms consider that the algorithm robustness of the worst time delay is obviously improved.
Efficiency of transmission：
Due to that network may be caused no longer to be connected to after arbitrary cut edge, part telephone net node to controller node
Time delay be infinity, cause the average delay of network that can not calculate and compare, thus select time delay inverse i.e. efficiency
Index come weigh controller dispose result quality.In SDN network, network transmission efficiency index (Transmission
Efficiency, abbreviation TE) it is defined as,Wherein, si=maxj∈c1/dij
(i ∈ V/C) indicates that nearest i.e. efficiency highest controller node j, d can be accessed in telephone net node ipqIndicate controller node
Number of edges between p and q on shortest path.First item ∑ in efficiency of transmission index TEi∈v/CsiIt is general in SDN network for describing
The efficiency of logical telephone net node access controller node, Section 2 are used for describing to exchange between any pair of controller node pair
Information is difficult to degree.SDN network efficiency of transmission index TE is bigger, shows switch-access controller resource and control in network
Exchange information is easier between device processed, and SDN network efficiency of transmission is higher.
Fig. 9 describes TE indexs as the worst scene moves down the increase of flash trimming number and downward trend.As shown in Figure 9 when
Break successively in OS3E networks 3 to 15 sides when, the efficiency of k-RCM algorithms is constantly in peak, i.e. efficiency of transmission is maximum.Thus
It is found that whether for the controller coverage rate index of SDN network, or for the transmission availability index of controller, and
Load balancing index, the k-RCM approximate datas with low computational complexity can obtain higher controller optimization configuration
Ability.
In another embodiment, in US Carrier as shown in Figure 10, it is assumed that given control coverage rate Cpr=
85%.For k-RCM approximate datas, under the worst scene for removing side, to ensure to meet control coverage rate Cpr, obtain most
Excellent controller configuration node set V0={ 6,18,28,39,48,60,67,69,86,91,99,102,116,136 } load feelings
Condition is shown in that color mark different in Figure 10, the node of same color are managed by the same controller.In the same circumstances, it is based on
The optimal controller configuration node set V that branch-bound algorithm obtains0=2,10,18,28,40,48,60,70,81,91,99,
102,125,136 }.
Figure 11 describes control overlay node number as removal number of edges purpose increases and downward trend variation, by comparing
Branch-bound algorithm and k-RCM approximate datas mutation analysis are it is found that have the k-RCM algorithms of low computational complexity to improving SDN
The advantage of network-based control overlay node number is clearly.Figure 12 describes efficiency of transmission TE indexs as removal number of edges purpose increases
Add and downward trend, we obtain controlling transmission efficiency of the k-RCM algorithms with low computational complexity to improvement SDN network
It is with the obvious advantage, k-RCM algorithms can not only effectively improve control overlay node number of the SDN network in link failure, also can
Significantly improve controller validity, while there is load balance process ability.
To sum up, it can analyze and obtain from Fig. 4 a to Figure 12, in removing SDN network under the worst situation on side, based on limited
The k-RCM algorithms of distributing rationally of SDN network controller can effectively improve control covering of the SDN network in face of link fails when
Rate and control effectiveness index, are capable of the biological treatability of lifting controller service, and negative when can solve no link fails
Carry equalization problem.In addition, by comparative analysis it is found that whether to controlling the influence of overlay node number, or to TE indexs
It influences, the k-RCM algorithms with low computational complexity are compared with the branch-bound algorithm with high computational complexity, k-
RCM algorithms equally there is higher SDN network controller optimization to dispose ability.
Although embodiment of the present invention is described above in association with attached drawing, the invention is not limited in above-mentioned
Specific embodiments and applications field, above-mentioned specific embodiment are only schematical, directiveness, rather than restricted
's.Those skilled in the art are under the enlightenment of this specification and in the range for not departing from the claims in the present invention and being protected
In the case of, a variety of forms can also be made, these belong to the row of protection of the invention.
Claims (10)
1. the multi-controller Optimization deployment method of robust, includes the following steps in a kind of software defined network：
S100：Analyze network topology structure
S101：Network topology structure is analyzed, the multi-controller deployment issue in software defined network SDN is abstracted and turns to a base
In the graph theoretic problem of undirected graph；
S102：It proposes a kind of new measurement, coverage rate index is controlled, for indicating that the interchanger sum that controller can be accessed accounts for
The ratio of all interchanger sums in network；
S200：Build Related Mathematical Models
Give a undirected graph G (V, L) and controller coverage rate Cpr, the node placed by reasonable deployment controller
Collect C so that the undirected graph is labeled as under the situation for arbitrarily removing k sideBoth met to
Fixed controller coverage rate Cpr, and network transmission efficiency and load balancing can be taken into account, wherein V indicates interchanger set, in nothing
Indicated into network with node, and n indicates the number of node, i.e. n=| V |, L indicates the link set of connection interchanger,
It is indicated with side in undirected graph, C indicates controller node set, i.e., the telephone net node set that controller is connected, in nothing
The scene set that any limit is removed into network G, is denoted as S, removes the situation set on k side, is denoted as sk,Remove k
The worst scene on side is denoted as sb, sb∈sk,Indicate the link set of the arbitrary connection interchanger for removing k side,It indicates
Remove the link set of the connection interchanger on k side under the worst situation, k be positive integer and 1≤k≤| L |；
S300：The selection and solution of approximate data
The k-RCM near-optimization algorithms for taking into account network delay and the robust of load balancing are chosen, k-RCM algorithms include mainly the worst
Scene moves down the GN algorithms of flash trimming, is denoted as k-RCM-GN, and Dual Approximate is denoted as k-RCM-DOLP and remote interchanger section
Point extraction algorithm, is denoted as k-RCM-outliers.
2. the method according to claim 1, which is characterized in that preferred, specific modeling is as follows：Object function：
Constraints：
Wherein：
n：Indicate the quantity of SDN network interior joint；
Sk：Indicate all situation set on k side of removal；
sk：Indicate a kind of situation on k side of removal, sk∈Sk；
sb：The corresponding the worst scene for removing k side；
Scene skThe probability of appearance,
fj：The cost of controller is placed on node j；
In scene skUnder, general switch node i is connected to the time delay that controller places node j；
cpr：Indicate given SDN controller coverage rates；
xj：It is a 0-1 variable, xj=1 indicates that node j places controller by selection, is otherwise 0；
It is a 0-1 variable, indicates node i in scene skIn whether node j covering placed by controller, if it is,Otherwise
It is a 0-1 variable, indicates node i in scene skIn whether be remote telephone net node, if it is,Otherwise
I, j indicate any one interchanger, i.e. i in interchanger set, j ∈ V.
3. method according to claim 2, which is characterized in that assuming that the worst cut edge scene sbProbability of happeningThen exist
It removes to model under the worst scene on k side and be updated to：
Object function：
Constraints：
4. method according to claim 3, which is characterized in that solve above-mentioned updated model, constrain 0-1 integers
Relaxation, modeling are updated to：
Object function：
Constraints：
5. method according to claim 4, which is characterized in that by above-mentioned constraints
It replaces with
6. method according to claim 5, which is characterized in that it is corresponding to establish its for the dual program for utilizing linear programming problem
Dual program model is：
Object function：
max∑i∈Vαi+n(α-1)q (L4-1)
Constraints：
∑j∈Vβij≤fj j∈V (L4-2)
Wherein, dual variable βijIndicate that the controller node j that telephone net node i connects it places cost fjContribution margin；It is right
Even variable αiIndicate the wastage in bulk or weight cost of telephone net node i, wastage in bulk or weight cost includes the time delay that interchanger i connects its controller j
cijCost f is placed with the controller that interchanger i connects itjContribution margin βij；Dual variable q indicates interchanger wastage in bulk or weight cost
αiMaximum value.
7. method according to claim 6, which is characterized in that the k-RCM near-optimizations algorithm in the step S300 is specifically such as
Under：
S301：Given undirected graph G (V, L) and controller coverage rate Cpr；
S302：Assuming that the maximum value that the controller in the known optimal deployment scheme of controller places cost is f ', f is changedj：Work as fj
When > f ', f is enabledj=∞, otherwise fjIt is constant, wherein fjIndicate the cost of placement controller on node j；
S303：Based on k-RCM-DOLP algorithms, remote node is extracted, remote set of node is deleted in G；
S304：Based on k-RCM-GN algorithms, the worst disconnected k side situation is found；
S305：Based on k-RCM-DOLP algorithms, the load table that controller places node set and each controller is generated；
S306：It is arranged to make remote node to be controlled by the controller of remote node and is connected thereto the controller of expense minimum.
8. method according to claim 7, which is characterized in that the k-RCM-GN algorithms are specific as follows：
A) initial setting up, num=0,
B) in given undirected graph it is all while by while betweenness center be ranked up, side is then deleted in undirected graph
The maximum side l of betweenness center index；
C) num=num+1 is corrected,
D) check whether num is equal to k, if num<K, then repeatedly step b and step c；If num=k thens follow the steps e；
E) collection on the k side that the worst scene of SDN network interrupts is combined into
9. method according to claim 7, which is characterized in that the k-RCM-DOLP algorithms are specific as follows：
a)αi=0, βij=0, mark all αiIt can increase, βijIt can not increase；
B) to increasable αiThere is αi=αi+1；
C) meet α with the presence or absence of (i, j)i=Cij；And if so, whether decision node j has been selected as the section for placing controller
Point, if it is flag node i controlled by node j；If otherwise label (i, j) connection side is tight, and βijIt is recycled in next time
In can increase；If there is no then executing next step；
D) to increasable βijThere is βij=βij+1；
E) meet restrictive condition ∑ with the presence or absence of node ji∈Vβij=fj(j∈V)；And if so, from restrictive condition ∑ is meti∈V
βij=fjDevice places node to the node j ' for selecting the attachable network equipment most in the node of (j ∈ V) in order to control；If do not deposited
Thening follow the steps g；
F) label can be by the network equipment of j ' controls, and βijAnd αiDo not increasing, i ∈ can be by the network equipment of j ' controls；
G) whether all nodes have been selected, if then terminating this algorithm, step b is executed if otherwise redirecting.
10. method according to claim 7, which is characterized in that the k-RCM-outliers algorithms are specific as follows：
A) assume that the maximum value that the controller in the optimal deployment scheme of known controller places cost is f ', change fj：Work as fj> f
' when, enable fj=∞, otherwise fjIt is constant, wherein fjIndicate the cost of placement controller on node j；
B) run k-RCM-DOLP algorithms, when have in network be less than (1-Cpr) * | V | a node be not determined as controller or
When the network equipment that person has been controlled, then k-RCM-DOLP algorithms are terminated, remaining node is then the remote node being selected；
C) the remote node is connected on the controller of nearest not load saturation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810518568.0A CN108777636B (en) | 2018-05-25 | 2018-05-25 | A kind of multi-controller Optimization deployment method of robust in software defined network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810518568.0A CN108777636B (en) | 2018-05-25 | 2018-05-25 | A kind of multi-controller Optimization deployment method of robust in software defined network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108777636A true CN108777636A (en) | 2018-11-09 |
| CN108777636B CN108777636B (en) | 2019-07-02 |
Family
ID=64027640
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810518568.0A Active CN108777636B (en) | 2018-05-25 | 2018-05-25 | A kind of multi-controller Optimization deployment method of robust in software defined network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108777636B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109921991A (en) * | 2019-01-14 | 2019-06-21 | 湘潭大学 | A kind of SDN controller portion arranging method based on Dinkelbach |
| CN110022271A (en) * | 2019-04-16 | 2019-07-16 | 中国人民解放军国防科技大学 | Robustness verification method and device for distributed control plane in software defined network |
| CN110233752A (en) * | 2019-05-28 | 2019-09-13 | 中国人民解放军战略支援部队信息工程大学 | A kind of the controller robust dispositions method and device of attack resistance |
| CN110830570A (en) * | 2019-11-01 | 2020-02-21 | 陕西师范大学 | Resource equalization deployment method for robust finite controller in software defined network |
| CN113347514A (en) * | 2021-06-22 | 2021-09-03 | 重庆邮电大学 | Software defined optical network controller deployment method based on multi-path survivability protection |
| CN114355775A (en) * | 2021-12-29 | 2022-04-15 | 航天科工网络信息发展有限公司 | Multi-controller deployment method and system based on SDN (software defined network) and deep reinforcement learning |
| US20220368615A1 (en) * | 2021-05-12 | 2022-11-17 | Vmware, Inc. | Agentless method to automatically detect low latency groups in containerized infrastructures |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104065590A (en) * | 2014-07-16 | 2014-09-24 | 福州大学 | Controller placement method for software-defined network |
| CN104065509A (en) * | 2014-07-24 | 2014-09-24 | 大连理工大学 | SDN multi-controller deployment method for reducing management load overhead |
| CN104410528A (en) * | 2014-12-09 | 2015-03-11 | 中国人民解放军国防科学技术大学 | Method for deploying minimum fault-tolerant coverage of controller based on software defined data center network |
| WO2015105987A1 (en) * | 2014-01-10 | 2015-07-16 | Huawei Technologies Co., Ltd. | System and method for zoning in software defined networks |
| CN105978740A (en) * | 2016-07-18 | 2016-09-28 | 北京邮电大学 | Method for deploying controller in software defined network |
| US20160323144A1 (en) * | 2015-04-29 | 2016-11-03 | Futurewei Technologies, Inc. | Traffic-driven network controller placement in software-defined networks |
| CN106100876A (en) * | 2016-06-03 | 2016-11-09 | 中国电子科技集团公司第三十研究所 | A kind of SDN middle controller dispositions method, path calculation method and system thereof |
| CN106452897A (en) * | 2016-10-26 | 2017-02-22 | 广东技术师范学院 | Method for placing controllers of software defined network |
| CN107171838A (en) * | 2017-05-18 | 2017-09-15 | 陕西师范大学 | It is a kind of that method for optimizing is reconstructed based on the Web content that limited content is backed up |
| CN107204874A (en) * | 2017-05-09 | 2017-09-26 | 天津大学 | Ensure the minimum SDN multi-controller dispositions method of time delay |
| CN107276662A (en) * | 2017-07-27 | 2017-10-20 | 大连大学 | A kind of software definition Information Network multi-controller dynamic deployment method |
-
2018
- 2018-05-25 CN CN201810518568.0A patent/CN108777636B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015105987A1 (en) * | 2014-01-10 | 2015-07-16 | Huawei Technologies Co., Ltd. | System and method for zoning in software defined networks |
| CN105900403A (en) * | 2014-01-10 | 2016-08-24 | 华为技术有限公司 | System and method for zoning in software defined networks |
| CN104065590A (en) * | 2014-07-16 | 2014-09-24 | 福州大学 | Controller placement method for software-defined network |
| CN104065509A (en) * | 2014-07-24 | 2014-09-24 | 大连理工大学 | SDN multi-controller deployment method for reducing management load overhead |
| CN104410528A (en) * | 2014-12-09 | 2015-03-11 | 中国人民解放军国防科学技术大学 | Method for deploying minimum fault-tolerant coverage of controller based on software defined data center network |
| US20160323144A1 (en) * | 2015-04-29 | 2016-11-03 | Futurewei Technologies, Inc. | Traffic-driven network controller placement in software-defined networks |
| CN106100876A (en) * | 2016-06-03 | 2016-11-09 | 中国电子科技集团公司第三十研究所 | A kind of SDN middle controller dispositions method, path calculation method and system thereof |
| CN105978740A (en) * | 2016-07-18 | 2016-09-28 | 北京邮电大学 | Method for deploying controller in software defined network |
| CN106452897A (en) * | 2016-10-26 | 2017-02-22 | 广东技术师范学院 | Method for placing controllers of software defined network |
| CN107204874A (en) * | 2017-05-09 | 2017-09-26 | 天津大学 | Ensure the minimum SDN multi-controller dispositions method of time delay |
| CN107171838A (en) * | 2017-05-18 | 2017-09-15 | 陕西师范大学 | It is a kind of that method for optimizing is reconstructed based on the Web content that limited content is backed up |
| CN107276662A (en) * | 2017-07-27 | 2017-10-20 | 大连大学 | A kind of software definition Information Network multi-controller dynamic deployment method |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109921991B (en) * | 2019-01-14 | 2021-06-04 | 湘潭大学 | SDN controller deployment method based on Dinkelbach |
| CN109921991A (en) * | 2019-01-14 | 2019-06-21 | 湘潭大学 | A kind of SDN controller portion arranging method based on Dinkelbach |
| CN110022271A (en) * | 2019-04-16 | 2019-07-16 | 中国人民解放军国防科技大学 | Robustness verification method and device for distributed control plane in software defined network |
| CN110022271B (en) * | 2019-04-16 | 2021-06-08 | 中国人民解放军国防科技大学 | Robustness verification method and device for distributed control plane in software defined network |
| CN110233752A (en) * | 2019-05-28 | 2019-09-13 | 中国人民解放军战略支援部队信息工程大学 | A kind of the controller robust dispositions method and device of attack resistance |
| CN110233752B (en) * | 2019-05-28 | 2021-11-09 | 中国人民解放军战略支援部队信息工程大学 | Robust deployment method and device for anti-attack controller |
| CN110830570A (en) * | 2019-11-01 | 2020-02-21 | 陕西师范大学 | Resource equalization deployment method for robust finite controller in software defined network |
| CN110830570B (en) * | 2019-11-01 | 2022-02-01 | 陕西师范大学 | Resource equalization deployment method for robust finite controller in software defined network |
| US20220368615A1 (en) * | 2021-05-12 | 2022-11-17 | Vmware, Inc. | Agentless method to automatically detect low latency groups in containerized infrastructures |
| US11729080B2 (en) * | 2021-05-12 | 2023-08-15 | Vmware, Inc. | Agentless method to automatically detect low latency groups in containerized infrastructures |
| CN113347514A (en) * | 2021-06-22 | 2021-09-03 | 重庆邮电大学 | Software defined optical network controller deployment method based on multi-path survivability protection |
| WO2022267350A1 (en) * | 2021-06-22 | 2022-12-29 | 重庆邮电大学 | Software defined optical network controller deployment method based on multi-path survivability protection |
| CN114355775A (en) * | 2021-12-29 | 2022-04-15 | 航天科工网络信息发展有限公司 | Multi-controller deployment method and system based on SDN (software defined network) and deep reinforcement learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108777636B (en) | 2019-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108777636B (en) | A kind of multi-controller Optimization deployment method of robust in software defined network | |
| Gouareb et al. | Virtual network functions routing and placement for edge cloud latency minimization | |
| CN108762768B (en) | Intelligent network service deployment method and system | |
| CN105191214B (en) | The network bandwidth allocations rented in system for cloud computing more | |
| US9301026B2 (en) | Affinity modeling in a data center network | |
| CN106464530B (en) | The method and system for the dynamic dispatching that network updates | |
| EP2774048B1 (en) | Affinity modeling in a data center network | |
| CN109842565A (en) | Interface current-limiting method, device, electronic equipment and storage medium | |
| CN104202264A (en) | Carrying resource allocation method for clouded data center network, device and system | |
| CN104125091A (en) | Automatic deployment method and device of network configuration | |
| CN103152197B (en) | Rule set layout processing method, device and cluster data system | |
| CN106095035A (en) | Server system and computer implemented method thereof and computer-readable recording medium | |
| CN106464531A (en) | Proactive handling of network faults | |
| CN113708972A (en) | Service function chain deployment method and device, electronic equipment and storage medium | |
| CN112822192B (en) | User-demand-oriented safety function service network system and implementation method thereof | |
| CN106162639A (en) | SDN wireless network management platform based on Floodlight and authentication method | |
| CN110830570B (en) | Resource equalization deployment method for robust finite controller in software defined network | |
| CN109479059A (en) | The system and method for transport layer class letter and isolation for container flow | |
| CN108092895A (en) | A kind of software defined network joint route selection and network function dispositions method | |
| CN109412963A (en) | A kind of service function chain dispositions method split based on stream | |
| Zhong et al. | Service function chain orchestration across multiple clouds | |
| Chaudhary et al. | PARC: Placement availability resilient controller scheme for software-defined datacenters | |
| CN113259175B (en) | Security service and function service combined arrangement method in edge computing environment | |
| Khatiri et al. | Balanced resource allocation for VNF service chain provisioning in inter-datacenter elastic optical networks | |
| CN113490279B (en) | Network slice configuration 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 |