CN109921991A - A kind of SDN controller portion arranging method based on Dinkelbach - Google Patents
A kind of SDN controller portion arranging method based on Dinkelbach Download PDFInfo
- Publication number
- CN109921991A CN109921991A CN201910031586.0A CN201910031586A CN109921991A CN 109921991 A CN109921991 A CN 109921991A CN 201910031586 A CN201910031586 A CN 201910031586A CN 109921991 A CN109921991 A CN 109921991A
- Authority
- CN
- China
- Prior art keywords
- controller
- domain
- deployment
- node
- optimal
- 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
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
- Telephonic Communication Services (AREA)
Abstract
The present invention proposes a kind of SDN controller portion arranging method based on Dinkelbach.SDN network topological structure is abstracted as non-directed graph by obtaining telephone net node information first, all nodes constitute female domain, then by judging that controller capacity conditions carry out a point domain, it using Dinkelbach method iterative search and saves domain internal controller and disposes optimal node location, finally determine the deployed position and quantity of whole controllers under the topological structure.The present invention adjusts controller loading condition according to the size in domain dynamic, and the deployment of load balancing between domain may be implemented as a result, increasing network flexibility, greatly improve the utilization rate of SDN controller, in addition, Dinkelbach method improves rate of convergence, the different network topology structure of scale can be suitably used for.
Description
Technical field
The invention belongs to software defined network fields, and in particular to a kind of dispositions method of software defined network controller.
Background technique
Software defined network (Software Defined Network, SDN) passes through decoupling control plane and data plane,
It supports the control of centralization network state, realizes that bottom-layer network infrastructure to the transparent of upper layer application, provides flexibly for network
Open interface and software programmability, scheduling on demand and the distribution of Internet resources can be supported, so that effective solution is current
The resource expansion that the network system faces is obstructed, and networking flexibility is poor, and emerging service demand is difficult to the problems such as meeting.
With the expansion of network topology structure and the increase of network flow, single SDN controller when managing network by
Many-sided limitation such as scalability, reliability, it is difficult to ensure that the network quality under catenet topological structure.Therefore for given
Network topology structure, the deployed position of multiple SDN controllers is managed be realize network quick response effective means.
In order to solve SDN controller deployment issue, time delay expense, load, capacity limit, node phase in research network structure can be passed through
Deployment issue is optimized like factors such as degree.
Summary of the invention
The invention proposes a kind of SDN controller portion arranging method based on Dinkelbach.Dinkelbach is that one kind changes
Into binary search algorithm, it is higher compared to dichotomy rate of convergence.
The deployment of this SDN controller at least the following steps are included:
Step 1 obtains telephone net node information and SDN network topological structure is abstracted as non-directed graph.Obtain interchanger section
Point average message traffic volume and average message delivery rate, all nodes constitute female domain.Female current SDN network topology knot of domain representation
It include the domain of all telephone net nodes under structure, other are subdomain by the domain that female domain divides.Due to when disposing SDN controller
It is typically chosen and SDN controller is deployed on some telephone net node, therefore will hereafter be generated in subdomain any in iterative process
Controller dispose optimal node and be known as disposing optimal node, will eventually determine that the node of deployment controller is known as controller portion
Affix one's name to node.Initialization controller disposes nodal information, including location information and quantity information.
Step 2 judges whether domain meets controller capacity conditions.If meeting controller capacity conditions, enter step
4, otherwise enter step 3.
Step 3, select to dispose in domain according to domain expense situation optimal node and according to dispose optimal node be associated with situation into
Row divides domain, and record disposes optimal nodal information.Specifically includes the following steps:
1) cancel the optimal node of deployment that last iteration generates and delete its location information, controller disposes node total number
Add 1.
2) optional 2 telephone net nodes generate all possible deployment scenario in domain as controller node.
3) the time delay expense of every kind of deployment scenario is calculated.
4) the telephone net node concentration degree of every kind of deployment scenario is calculated.
5) deployment benefit is calculated, optimal controller deployment scenario is searched for by Dinkelbach, obtains 2 optimal sections of deployment
Point.
6) domain is divided into 2 subdomains according to the association situation of 2 optimal nodes of deployment and records 2 optimal nodes of deployment
Location information.
7) 2 subdomains are returned into step 2.
Step 4 judges whether domain is female domain, and is obtained in different ways according to judging result and save controller deployment section
Point information.Include following two situation:
When domain is female domain, 1 optimal controller of search disposes node directly in female domain, saves optimal controller deployment
Node location information, controller deployment node total number save as 1.Enter step 5.
When domain is subdomain, pass through 6 in step 3) location information for disposing optimal node of final subdomain is obtained, and incite somebody to action
It is saved as the location information of controller deployment node, is terminated current subdomain and is divided.It waits whole subdomains to divide to complete
Afterwards, the quantity information of controller deployment node is saved.Enter step 5.
The controller deployment node location information and quantity information that step 5, output save, terminate deployment.
The invention has the advantages that:
1. pair domain carries out two points, dynamic adjusts controller loading condition, can achieve the purpose of load balancing, mentions
High network processes efficiency, the utilization rate for adding somewhat to the flexibility of network, improving SDN controller.
Dinkelbach algorithm is used when 2. searching for optimal deployment scenario, convergence is accelerated, as network size expands this
The advantage of kind method can be more obvious.
Detailed description of the invention
Fig. 1 is the flow chart of invention.
Fig. 2 is female domain by SDN network topological structure.
Fig. 3 is optimal deployment scenario in domain.
Fig. 4 is the result in point domain for the first time.
Fig. 5 is the result in point domain for the second time.
Fig. 6 is the search result that optimal controller is directly carried out to female domain.
Specific embodiment
As shown in Figure 1, specific steps of the invention are as follows:
Step 1 obtains telephone net node information and SDN network topological structure is abstracted as a non-directed graph G (V, L), and V is
The set of interchanger in network, V={ v1,v2,…,vn, n is the quantity of telephone net node in SDN network structure, and L is interchanger
Between link range matrix, LijIndicate telephone net node viTo telephone net node vjLink range.Obtain each interchanger
The average message transmission rate t of nodeiAnd average message traffic volume ui, all female domains of telephone net nodes composition.Initialization controller
Nodal information is disposed, sets empty for controller deployment node location information, quantity is set as 0.Assuming that the interchanger used is equal
For OpenFlow interchanger.Network topology structure shows that wherein Fig. 2 to Fig. 5 belongs to identical network topology knot by Fig. 2 to Fig. 6
Structure, Fig. 6 belong to the lesser network topology structure of another scale.
Step 2 judges whether domain meets controller capacity conditions.Assuming that every controller is D to the capacity of size of message,
Domain is SVnow.Controller capacity conditions areI.e. controller capacity is not less than domain SVnowInterior size of message
Total amount.If meeting controller capacity conditions, 4 are entered step, otherwise enters step 3.
Step 3 selects to dispose optimal node in optimal domain and is associated with feelings according to optimal node is disposed according to domain expense situation
Condition carries out a point domain, and record disposes optimal nodal information.The following steps are included:
1) cancel the optimal node of deployment that last iteration generates and delete its location information, controller disposes node total number
Add 1.
2) optional 2 telephone net nodes generate a kind of possible controller deployment scenario, similarly as controller node
It can produce deployment scenario all in domain.Assuming that domain is SVnow, can remember SVnowThe collection of interior all possible deployment scenario is combined into
{X1,X2,…,Xm, m is possible deployment scenario total amount.
3) the time delay expense of every kind of deployment scenario is calculated.
Define deployment scenario Xd(d=1,2 ... time delay expense m) are as follows: Dd=Dtu+Dtp。Dtu、DtpRespectively indicate deployment
Situation Xd(d=1,2 ... m) lower 2 controller node vu、vpExpense.Controller node vu、vpThe expense representation method phase at place
Together, by taking the expense of controller node vu as an example, it is assumed that each controller is H to the processing speed of message, it is known that l is to hand in domain
It changes planes number of nodes:
DtuFor controller node vuOverall delay expense.DpiuIndicate telephone net node viWith controller node vuBetween biography
Sowing time prolongs, wherein LiuFor telephone net node viTo controller node vuControl linkage length, LiuThe chain in non-directed graph G can be passed through
Road distance matrix L is obtained, and c indicates Electromagnetic Wave Propagation rate, and controller is deployed in certain due to being typically chosen when deployment controller
On a telephone net node, therefore controller node v is not calculateduTo the propagation delay of this controller, i.e. Dpuu=0；DtriIt indicates to hand over
Change planes node viThe transmission delay at place, uiAnd tiIt is telephone net node v respectivelyiAverage message traffic volume and average message send
Rate；DhiIndicate telephone net node viProcessing delay of the transmitted message at controller.yiuIndicate telephone net node viWith
Controller node vuAssociation situation, yiu=1 indicates telephone net node viWith controller node vuIt is associated, yiu=0 indicates to hand over
Change planes node viWith controller node vuIt is unrelated.
Determine that telephone net node and the associated method of controller node are as follows: it is required that in domain 2 controller nodes to both
Associated telephone net node overhead is minimum, establishes such as drag:
DiuWith DipRespectively indicate telephone net node viTo controller node vuWith vpNode between expense, wherein DiuIt can express
Are as follows: Diu=2Dpiu+Dtri+Dhi, DipWith DiuExpression way is consistent.Indicate that controller is negative
Carry equilibrium condition, i.e. controller node vuThe size of message and controller node v of processingpThe absolute value of the size of message difference of processing exists
The accounting of domain inside information amount is less than δ (0 < δ < 1), and δ indicates that the maximum load difference allowed accounts for the ratio of domain inside information amount, δ
Load difference is smaller between lower representative domain, and controller load balance is better.yiu∈{0,1}And yip=
1-yiuIndicate that each telephone net node can only be associated with 1 in 2 controller nodes.yuu=1, ypp=1 indicates to control
Device node is associated with itself.Situation is associated with by what above-mentioned model can obtain telephone net node and controller node.
4) the telephone net node concentration degree of every kind of deployment scenario is calculated.Define deployment scenario Xd(d=1,2 ... exchange m)
Machine node concentration degree are as follows:
What denominator indicated is 2 controller nodes to associated the sum of the linkage length of telephone net node, Liu,LipPoint
It Biao Shi not telephone net node viTo controller node vu,vpLinkage length.Since telephone net node quantity l does not change in domain, point
Female bigger expression controller node vuWith vpIt is bigger to the sum of telephone net node linkage length associated by the two, it will lead to exchange
Machine node concentration degree CrdDecline.
5) deployment benefit is calculated, optimal deployment scenario is searched for by Dinkelbach, obtains 2 optimal nodes of deployment.
Deployment benefit is expressed asDdFor deployment scenario XdUnder time delay expense, CrdIt is deployment scenario XdLower telephone net node
Concentration degree.Therefore for DdBigger and CrdSmaller deployment scenario disposes benefitLower, objective function is set asIt is searched for using Dinkelbach2 controller nodes corresponding to obtained optimal deployment scenario
Optimal node is as disposed, as a result as shown in figure 3, indicating to have determined that 2 optimal nodes of deployment in domain.
6) domain is divided into 2 subdomains by the association situation according to 2 optimal nodes of deployment.If domain is SVnow, 2 portions of acquisition
Affixing one's name to optimal node is vuAnd vp, then by SVnowIt is divided into 2 subdomain SVuWith SVp.It then will be with vuAnd vpAssociated node difference
It is divided into SVuWith SVpIn, such as node viAssociation situation be yiu=1, yip=0, then by viIt is divided into subdomain SVuIt is interior.For the first time
Iteration result is as shown in figure 4, be divided into 2 subdomains for female domain, and second of iteration result is as shown in figure 5,2 subdomains are divided into
4 subdomains.Record disposes optimal node location information.
7) 2 subdomains are returned into step 2.
Step 4 judges whether domain is female domain, and is obtained in different ways according to judging result and save controller deployment section
The location information and quantity information of point include following 2 kinds of situations:
1) if domain is female domain, it can learn that network topology structure scale is smaller, be disposed using 1 controller
Meet domain internal controller capacity requirement.1 optimal controller of search disposes node directly in female domain, and searching method is equally
Dinkelbach.Search objective function is changed toWhereinIndicate that only choosing 1 telephone net node makees
Deployment benefit when node is disposed for controller.Assuming that controller deployment node is vu,Indicate only deployment 1
Telephone net node concentration degree when a controller.Ds=Dtu, indicate time delay expense when only disposing 1 controller, DtuExpression is such as
Under:
Y at this timeiu≡ 1 indicates that controller deployment node and all telephone net nodes in female domain are associated with.Search for female domain most
Excellent controller disposes node, and example results are as shown in Figure 6.Save the location information of optimal controller deployment node, controller portion
Administration's number of nodes saves as 1, enters step 5.
2) if domain is subdomain, illustrates to have obtained final subdomain by one or many iteration, step can be passed through
6 in rapid 3) location information of the optimal node of deployment of final subdomain is obtained, and as the position of controller deployment node
Information is saved, and is terminated current subdomain and is divided.Whole subdomains are waited to divide 1 for obtaining and saving after the completion from step 3) it obtains
Controller dispose number of nodes information, enter step 5.
The controller deployment node location and quantity information that step 5, output save, terminate deployment.
Claims (5)
1. a kind of dispositions method of SDN controller, the method at least include the following steps:
Step 1: telephone net node information is obtained by SDN network topological structure and is abstracted as non-directed graph, it is average to obtain telephone net node
Message traffic volume and average message delivery rate, all nodes constitute female domain, and initialization controller disposes nodal information, including position
Confidence breath, quantity information；
Step 2: judging whether domain meets controller capacity conditions, is chosen whether to enter step three iteration according to result；
Step 3: it selects to dispose optimal node in domain according to domain expense situation and be divided according to optimal node association situation is disposed
Domain, record dispose optimal nodal information；
Step 4: judging whether domain is female domain, and is obtained in different ways according to judging result and save controller deployment node
Information；
Step 5: controller the deployment node location information, quantity information of preservation are exported, deployment is terminated.
2. the dispositions method of SDN controller according to claim 1, it is characterised in that the step 3 is according to domain expense feelings
Optimal node is disposed in condition selection domain and carries out a point domain according to optimal node association situation is disposed, and record disposes optimal nodal information
It at least includes the following steps:
Step 1: cancelling the optimal node of deployment that last iteration generates and delete its location information, controller disposes node total number
Add 1；
Step 2: optional 2 telephone net nodes generate all possible deployment scenario in domain as controller node；
Step 3: calculating the time delay expense of every kind of deployment scenario；
Step 4: calculating the telephone net node concentration degree of every kind of deployment scenario；
Step 5: calculating deployment benefit, optimal controller deployment scenario is searched for by Dinkelbach, obtain 2 optimal sections of deployment
Point；
Step 6: domain is divided into 2 subdomains and records 2 optimal nodes of deployment by the association situation according to 2 optimal nodes of deployment
Location information.
3. the dispositions method of SDN controller according to claim 1, it is characterised in that the controller capacity conditions are to change
The condition in generation, wherein controller capacity conditions beD indicates controller to the capacity of size of message, and domain is
SVnow, i.e., controller capacity is not less than domain SVnowInterior size of message total amount.
4. the dispositions method of SDN controller according to claim 1, it is characterised in that step 4 judges whether domain is female
Domain, and controller deployment nodal information is obtained according to judging result in different ways and specifically includes two kinds of situations:
Situation 1: when domain is female domain, optimal controller is directly searched in female domain and disposes node, save optimal controller deployment
Node location information, controller deployment number of nodes save as 1；
Situation 2: it when domain is subdomain, obtains iteration result and saves the controller deployment node location information of final subdomain, control
Device deployment number of nodes processed saves after the completion of whole domains divide.
5. the dispositions method of SDN controller according to claim 2, it is characterised in that the target letter of Dinkelbach search
Number isWhereinIndicate deployment benefit, wherein CrdThat indicate is deployment scenario XdLower telephone net node collection
Moderate, DdThat indicate is deployment scenario XdUnder time delay expense.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910031586.0A CN109921991B (en) | 2019-01-14 | 2019-01-14 | SDN controller deployment method based on Dinkelbach |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910031586.0A CN109921991B (en) | 2019-01-14 | 2019-01-14 | SDN controller deployment method based on Dinkelbach |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109921991A true CN109921991A (en) | 2019-06-21 |
| CN109921991B CN109921991B (en) | 2021-06-04 |
Family
ID=66960250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910031586.0A Active CN109921991B (en) | 2019-01-14 | 2019-01-14 | SDN controller deployment method based on Dinkelbach |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109921991B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111181792A (en) * | 2019-12-31 | 2020-05-19 | 中移（杭州）信息技术有限公司 | SDN controller deployment method and device based on network topology and electronic equipment |
| CN114374427A (en) * | 2022-01-19 | 2022-04-19 | 苏州全时空信息技术有限公司 | Dynamic deployment method and system of controller in software-defined medium and low orbit satellite network |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105515987A (en) * | 2015-12-02 | 2016-04-20 | 西安电子科技大学 | SDN framework based virtual optical network oriented mapping method |
| CN105656771A (en) * | 2014-11-12 | 2016-06-08 | 中国移动通信集团公司 | Service path determining method, device and system |
| US9596141B2 (en) * | 2013-03-15 | 2017-03-14 | Cisco Technology, Inc. | Representing software defined networks using a programmable graph model |
| US10057112B2 (en) * | 2016-03-24 | 2018-08-21 | Ca, Inc. | Fault detection of service chains in a SDN/NFV network environment |
| CN108551406A (en) * | 2018-04-24 | 2018-09-18 | 福州大学 | A kind of software definition mobile radio network channel division method based on load balancing |
| CN108777636A (en) * | 2018-05-25 | 2018-11-09 | 陕西师范大学 | A kind of multi-controller Optimization deployment method of robust in software defined network |
| CN108923979A (en) * | 2018-07-11 | 2018-11-30 | 重庆邮电大学 | Software defined network mapping method of virtual network |
-
2019
- 2019-01-14 CN CN201910031586.0A patent/CN109921991B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9596141B2 (en) * | 2013-03-15 | 2017-03-14 | Cisco Technology, Inc. | Representing software defined networks using a programmable graph model |
| CN105656771A (en) * | 2014-11-12 | 2016-06-08 | 中国移动通信集团公司 | Service path determining method, device and system |
| CN105515987A (en) * | 2015-12-02 | 2016-04-20 | 西安电子科技大学 | SDN framework based virtual optical network oriented mapping method |
| US10057112B2 (en) * | 2016-03-24 | 2018-08-21 | Ca, Inc. | Fault detection of service chains in a SDN/NFV network environment |
| CN108551406A (en) * | 2018-04-24 | 2018-09-18 | 福州大学 | A kind of software definition mobile radio network channel division method based on load balancing |
| CN108777636A (en) * | 2018-05-25 | 2018-11-09 | 陕西师范大学 | A kind of multi-controller Optimization deployment method of robust in software defined network |
| CN108923979A (en) * | 2018-07-11 | 2018-11-30 | 重庆邮电大学 | Software defined network mapping method of virtual network |
Non-Patent Citations (1)
| Title |
|---|
| 蒋红艳: "《基于流量监控的网络性能优化关键技术研究》", 《中国博士学位论文全文数据库信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111181792A (en) * | 2019-12-31 | 2020-05-19 | 中移（杭州）信息技术有限公司 | SDN controller deployment method and device based on network topology and electronic equipment |
| CN111181792B (en) * | 2019-12-31 | 2023-04-07 | 中移（杭州）信息技术有限公司 | SDN controller deployment method and device based on network topology and electronic equipment |
| CN114374427A (en) * | 2022-01-19 | 2022-04-19 | 苏州全时空信息技术有限公司 | Dynamic deployment method and system of controller in software-defined medium and low orbit satellite network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109921991B (en) | 2021-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112020103B (en) | Content cache deployment method in mobile edge cloud | |
| CN105515987B (en) | A kind of mapping method based on SDN framework Virtual optical-fiber networks | |
| Wu et al. | Modeling and analysis of communication networks in multicluster systems under spatio-temporal bursty traffic | |
| CN109347657B (en) | Method for constructing virtual data domain of scientific and technological service under SDN mode | |
| CN111988225A (en) | Multi-path routing method based on reinforcement learning and transfer learning | |
| CN109379230B (en) | Service function chain deployment method based on breadth-first search | |
| CN109951391B (en) | Network path rapid calculation method based on multi-QoS constraint | |
| CN109921991A (en) | A kind of SDN controller portion arranging method based on Dinkelbach | |
| CN109919532A (en) | Logistics node determination method and device | |
| CN105677447A (en) | Clustering-based delay bandwidth minimization virtual machine deployment method in distributed cloud | |
| CN114785692B (en) | Communication network flow balancing method and device for aggregation regulation of virtual power plants | |
| CN111460277B (en) | Personalized recommendation method based on mobile social network tree-shaped transmission path | |
| CN107426315A (en) | A kind of improved method of the distributed cache system Memcached based on BP neural network | |
| CN114513449A (en) | Method and system for optimizing routing in domain | |
| Xie et al. | Mesh-of-Torus: a new topology for server-centric data center networks | |
| Wang et al. | An efficient and reliable service customized routing mechanism based on deep learning in IPv6 network | |
| CN106789147A (en) | A kind of flow analysis method and device | |
| CN116132354B (en) | Unmanned aerial vehicle cluster networking transmission path optimization method and system | |
| CN117478518A (en) | Map convolution neural network assisted big data storage forwarding scheduling method | |
| CN108366015A (en) | Routing algorithm for software defined network | |
| Duato | Improving the efficiency of virtual channels with time-dependent selection functions | |
| CN104348695A (en) | Artificial immune system-based virtual network mapping method and system thereof | |
| CN111131064A (en) | Multicast stream scheduling method in data center network | |
| Modiano et al. | A simple analysis of average queueing delay in tree networks | |
| CN110138670B (en) | Load migration method based on dynamic path |
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 |