CN107241277A - Node method for annealing in SDN - Google Patents
Node method for annealing in SDN Download PDFInfo
- Publication number
- CN107241277A CN107241277A CN201710347930.8A CN201710347930A CN107241277A CN 107241277 A CN107241277 A CN 107241277A CN 201710347930 A CN201710347930 A CN 201710347930A CN 107241277 A CN107241277 A CN 107241277A
- Authority
- CN
- China
- Prior art keywords
- node
- variance
- now
- iterations
- class
- 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.)
- Pending
Links
Classifications
-
- 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
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/25—Routing or path finding in a switch fabric
- H04L49/253—Routing or path finding in a switch fabric using establishment or release of connections between ports
- H04L49/254—Centralised controller, i.e. arbitration or scheduling
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to next generation network control and management and software defined network field, to ensure the load balancing of controller.Node method for annealing in the present invention, SDN, step is as follows：Step 1：Preliminary classification is carried out to network according to improved K medoids algorithms, the tag set Q of classification results center point set P and each node is returned；Step 2：Iterations is set and the initial variance of whole network nodes is calculated；Step 3：All nodes of traversal, find all boundary nodes；Step 4：In all boundary nodes, look for a boundary node not operated to carry out reclassifying operation, given the class of node adjacent thereto；Step 5：Variance now is calculated, is compared with previous variance：Step 6：Judge now whether iterations reaches：If not up to iterations, return to step 4；Otherwise, classification results now are returned, are terminated.Present invention is mainly applied to software defined network occasion.
Description
Technical field
It it is a kind of novel be used for soft the invention belongs to next generation network control and management and software defined network field
Part defines network rationalization partition and controller Deployment Algorithm, is related to one kind in the case where time delay is minimum to software defined network
Network full wafer network carries out subregion, and ensures the controller Deployment Algorithm of each controller load balancing.Concretely relate in SDN
Node method for annealing.
Background technology
Software defined network (software-defined network, SDN) is a kind of new network architecture.Its master
Wanting thought is the separation of datum plane and control plane, so as to support the network of centralization to control.Under this configuration, interchanger
Only responsible data forwarding, control logic is provided by controller.
In widely used SDN schemes Openflow, data forwarding is in units of flowing, and a stream is according to network address net
The information such as network port and protocol type are defined.Multiple flow tables are store in Openflow interchangers to instruct turning for stream
Hair.When data flow enters Openflow interchangers, exchange opportunity forwards data according to the flow table item matched with the stream.If
There is no matching forwarding in the flow table of interchanger, then inquiry can be sent to controller.After controller makes a policy, then
New flow table item is issued to interchanger.This set mechanism allows manager more easily to manage whole network.
At the same time, because controller bears the work of whole network, the disposal ability and controller of controller with
The time delay communicated between interchanger has important influence to the performance of whole network.However, the capacity of controller is also limited
, with the fast development of SDN, a controller has not born the business of a huge SDN.This will
SDN is asked to be distributed multiple controllers of formula to share the pressure of whole network.
Therefore, in order that SDN can normal work and more efficient, in the case of proof load in a balanced way,
Rational controller deployment is carried out to whole SDN.
Document [1] proposes the deployment issue of controller earliest, and measurement index is used as using average delay and maximum delay
Carry out the deployment issue of analyzer-controller, and it is solved using greedy algorithm.Document [2] proposes K-critical algorithms, root
Amount controller and deployed position needed for being calculated according to maximum allowable delay, but the document is also without the negative of consideration controller
Carry equalization problem.Controller deployment issue of the document [3] in this WAN is solved using spectral clustering, its algorithm has one
Fixed load balancing effect.Document [4] has used particle cluster algorithm to solve SDN controller deployment issue, considers simultaneously
Time delay and load balancing.
Simulated annealing [5] is the algorithm for solving optimization problem.It simulates the annealing in thermodynamic system
Process.It receives the solution of the current solution difference of a ratio with certain probability, and it is therefore possible to can jump out the optimal of this part
Solution, reaches the optimal solution of the overall situation.Inspired by the algorithm, the present invention proposes a kind of equal for solving the load of SDN controllers
The algorithm of weighing apparatus problem, and it is named as node annealing algorithm.
[1]HELLER B,SHERWOOD R,MCKEOWN N.The controller placement problem
[C]//Proceedings of the First Workshop on Hot Topics in Software Defined
Networks.ACM,2012:7-12
[2] Y, C,GARCIA A J.On the controller
placement for designing a distributed SDN control layer[C]//Networking
Conference,2014IFIP.IEEE,2014:1-9.
[3]Xiao P,Qu W,Qi H,et al.The SDN controller placement problem for
WAN[C]//Ieee/cic International Conference on Communications in China.IEEE,
2014:220-224.
[4]GAO C,WANG H,ZHU F,et al.A particle swarm optimization algorithm
for controller placement problem in software defined network[C]//
International Conference on Algorithms and Architectures for Parallel
Processing.Springer International Publishing,2015:44-54
[5]Bertsimas D,Tsitsiklis J.Simulated Annealing[J].Statistical
Science,1993,8(1):10-15。
The content of the invention
To overcome the deficiencies in the prior art, the present invention is directed to propose the node annealing algorithm in a kind of SDN, it is ensured that controller
Load balancing.The technical solution adopted by the present invention is that the node method for annealing in SDN, step is as follows：
Step 1：Preliminary classification is carried out to network according to improved K-medoids algorithms, classification results central point is returned
Set P and each node tag set Q；
Step 2：Iterations is set and the initial variance of whole network nodes is calculated；
Step 3：All nodes of traversal, find all boundary nodes；
Step 4：In all boundary nodes, a boundary node not operated is looked for carry out reclassifying operation, will
It gives the class of node adjacent thereto；
Step 5：Variance now is calculated, is compared with previous variance：
If now variance is less than previous variance, receive reclassifying in step 4；
Otherwise, do not receive to reclassify, repeat step 4；
Step 6：Judge now whether iterations reaches：
If not up to iterations, return to step 4；
Otherwise, classification results now are returned, are terminated.
Improved K-medoids algorithms, step is as follows：
Step 1：A node arbitrarily is selected as initial center point from N number of node；
Step 2：Central point is updated according to K-medoids algorithms；
Step 3：Node farthest apart from central point in each class is found, it is present in a set, selects the set
In to all central points apart from sum it is maximum o'clock be used as a new central point；
Step 4：Repeat step 2, judges whole network either with or without being divided into K class：
If being not separated into K class, repeat step 3 simultaneously continues；
If being divided into K class, classification results are returned to, terminate algorithm.
In an example, comprise the following steps that：
Step 1：Preliminary classification is carried out to network according to improved K-medoids algorithms, classification results are respectively central point
Set P and each point label set Q；
Step 2：Setting iterations is D, calculates the initial variance T of now whole network telephone net node number0；
Step 3：Institute in Q is traveled through a little, to find all boundary nodes, store it in T in set B1；
Step 4：Calculate now variance T1, the boundary node B inside a set B is taken at randomi, i=0,1 ..., N-1 will
BiLabel change into the label of adjacent inhomogeneity node, that is, the class for assigning to adjacent inhomogeneity node, calculate now variance T2；
Step 5：Calculate Δ T=T2-T1.If Δ T ＜ 0, receive BiLabel change, update Q；Otherwise, then B is not receivedi
The change of label, return to step 4；
Step 6：Judge whether to reach iterations D：
If not up to iterations D, return to step 4；
If having reached iterations D, P and Q now is returned, is terminated.
The features of the present invention and beneficial effect are：
Advantage of the present invention is mainly the problem of load balancing of the controller of consideration, can provide rational controller deployment knot
Really.
In the research disposed to SDN controllers, most of is all that the time delay using interchanger to controller is used as deployment mesh
Mark.This method has no problem in the case where controller capacity is very big.But once network becomes large-scale, controller
The problem of capacity is inevitable.This algorithm is solved using node annealing algorithm to SDN controller deployment issues, it is ensured that
The interchanger number of each subnet controller controls of SDN keeps balance.Solution procedure complexity is relatively low, and run time is extremely short.
In terms of solving result performance estimation, this algorithm measurement index of the nodes variance as two performances.As a result
It is illustrated in fig. 3 shown below, by taking U.S. Internet2Advanced Layer2Services topologys as an example, it is assumed that K=3, its interior joint
7,16,30 be respectively Centroid, and the nodes of each class are respectively 13,13,13, and variance is 0.
Brief description of the drawings：
Fig. 1 boundary nodes (1,5,6,7,8,9).
Fig. 2 algorithm flow charts.
(central point is 7,16,30 to classification results, and 0) variance is when Fig. 3 disposes 3 controllers.
Embodiment
The controller deployment issue although current existing scholar begins one's study, effective scheme is but or seldom.Especially
In terms of the load balancing of controller, rarely document can provide rational solution.The capacity of each controller in SDN
It is limited, once the load of controller has exceeded its capacity, whole network necessarily can not normal work.The purpose of the present invention exists
In proposing a kind of proof load SDN controllers Deployment Algorithm in a balanced way, it is ensured that controller is deployed in rational position.
The main process of this algorithm is, first to carrying out preliminary classification to whole network according to time delay.Calculate whole again
The nodes variance of each class of network.Then all nodes are traveled through, all boundary nodes are found (i.e. on the border of its own class
And the node being joined directly together with other classes, as shown in Figure 1), exist in a set.Some of set node is carried out
Reclassify.If the variance after reclassifying is less than original variance, receive this classification；Otherwise do not receive.Repeat always
Said process is until reaching iterations.
Concrete scheme is as follows：
A kind of node annealing algorithm in SDN：
It is primarily based on the minimum classification initial to whole network progress of time delay.Improved K-medoids algorithms are used,
Step is as follows：
Step 1：A node arbitrarily is selected as initial center point from N number of node.
Step 2：Central point is updated according to K-medoids algorithms.
Step 3：Node farthest apart from central point in each class is found, it is present in a set.Select the set
In to all central points apart from sum it is maximum o'clock be used as a new central point.
Step 4：Repeat step 2.Judge whole network either with or without being divided into K class：
If being not separated into K class, repeat step 3 simultaneously continues；
If being divided into K class, classification results are returned to, terminate algorithm.
Node annealing algorithm in SDN：
Step 1：Preliminary classification is carried out to network according to improved K-medoids algorithms, classification results are respectively P (centers
The set of point) and Q (set of each point label).
Step 2：Setting iterations is D, calculates the initial variance T of now whole network telephone net node number0。
Step 3：Institute in Q is traveled through a little, to find all boundary nodes, store it in T in set B1。
Step 4：Calculate now variance T1.The random boundary node B taken inside a set Bi, i=0,1 ..., N-1 will
BiLabel change into the label of adjacent inhomogeneity node, that is, the class for assigning to adjacent inhomogeneity node.Calculate now variance T2。
Step 5：Calculate Δ T=T2-T1.If Δ T ＜ 0, receive BiLabel change, update Q；Otherwise, then B is not receivedi
The change of label, return to step 4.
Step 6：Judge whether to reach iterations D：
If not up to iterations D, return to step 4；
If having reached iterations D, P (set of central point) and Q (set of each point label) now is returned, is terminated
Algorithm.
Claims (3)
1. the node method for annealing in a kind of SDN, it is characterized in that, step is as follows：
Step 1：Preliminary classification is carried out to network according to improved K-medoids algorithms, classification results center point set P is returned
With the tag set Q of each node；
Step 2：Iterations is set and the initial variance of whole network nodes is calculated；
Step 3：All nodes of traversal, find all boundary nodes；
Step 4：In all boundary nodes, a boundary node not operated is looked for carry out reclassifying operation, by its point
Give the class of its adjacent node；
Step 5：Variance now is calculated, is compared with previous variance：
If now variance is less than previous variance, receive reclassifying in step 4；
Otherwise, do not receive to reclassify, repeat step 4；
Step 6：Judge now whether iterations reaches：
If not up to iterations, return to step 4；
Otherwise, classification results now are returned, are terminated.
2. the node method for annealing in SDN as claimed in claim 1, it is characterized in that, improved K-medoids algorithms, step
It is as follows：
Step 1：A node arbitrarily is selected as initial center point from N number of node；
Step 2：Central point is updated according to K-medoids algorithms；
Step 3：Node farthest apart from central point in each class is found, it is present in a set, selects to arrive in the set
All central points apart from sum it is maximum o'clock be used as a new central point；
Step 4：Repeat step 2, judges whole network either with or without being divided into K class：
If being not separated into K class, repeat step 3 simultaneously continues；
If being divided into K class, classification results are returned to, terminate algorithm.
3. the node method for annealing in SDN as claimed in claim 1, it is characterized in that, in an example, specific steps are such as
Under：
Step 1：Preliminary classification is carried out to network according to improved K-medoids algorithms, classification results are respectively the collection of central point
Close the set Q of P and each point label；
Step 2：Setting iterations is D, calculates the initial variance T of now whole network telephone net node number0；
Step 3：Institute in Q is traveled through a little, to find all boundary nodes, store it in T in set B1；
Step 4：Calculate now variance T1, the boundary node B inside a set B is taken at randomi, i=0,1 ..., N-1, by Bi's
Label changes into the label of adjacent inhomogeneity node, that is, in the class for assigning to adjacent inhomogeneity node, calculates now variance T2；
Step 5：Calculate Δ T=T2-T1.If Δ T ＜ 0, receive BiLabel change, update Q；Otherwise, then B is not receivediLabel
Change, return to step 4；
Step 6：Judge whether to reach iterations D：
If not up to iterations D, return to step 4；
If having reached iterations D, P and Q now is returned, is terminated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710347930.8A CN107241277A (en) | 2017-05-17 | 2017-05-17 | Node method for annealing in SDN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710347930.8A CN107241277A (en) | 2017-05-17 | 2017-05-17 | Node method for annealing in SDN |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107241277A true CN107241277A (en) | 2017-10-10 |
Family
ID=59984503
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710347930.8A Pending CN107241277A (en) | 2017-05-17 | 2017-05-17 | Node method for annealing in SDN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107241277A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401196A (en) * | 2020-03-10 | 2020-07-10 | 珠海全志科技股份有限公司 | Method, computer device and computer readable storage medium for self-adaptive face clustering in limited space |
| CN113542121A (en) * | 2021-07-05 | 2021-10-22 | 浙江大学 | Load balancing routing method for tree data center link layer based on annealing method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104717276A (en) * | 2013-12-12 | 2015-06-17 | 国际商业机器公司 | Method and system for allocating data storage in network |
| WO2016060751A1 (en) * | 2014-10-13 | 2016-04-21 | Nec Laboratories America, Inc. | Network traffic flow management using machine learning |
| CN106341346A (en) * | 2016-09-08 | 2017-01-18 | 重庆邮电大学 | Routing algorithm of guaranteeing QoS in data center network based on SDN |
-
2017
- 2017-05-17 CN CN201710347930.8A patent/CN107241277A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104717276A (en) * | 2013-12-12 | 2015-06-17 | 国际商业机器公司 | Method and system for allocating data storage in network |
| WO2016060751A1 (en) * | 2014-10-13 | 2016-04-21 | Nec Laboratories America, Inc. | Network traffic flow management using machine learning |
| CN106341346A (en) * | 2016-09-08 | 2017-01-18 | 重庆邮电大学 | Routing algorithm of guaranteeing QoS in data center network based on SDN |
Non-Patent Citations (2)
| Title |
|---|
| 王艳娥: "《划分式聚类算法的初始化方法研究》", 《硕士学位论文》 * |
| 郭佳: "《一种SDN环境中的网络节点重要性排序算法》", 《硕士学位论文》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401196A (en) * | 2020-03-10 | 2020-07-10 | 珠海全志科技股份有限公司 | Method, computer device and computer readable storage medium for self-adaptive face clustering in limited space |
| CN113542121A (en) * | 2021-07-05 | 2021-10-22 | 浙江大学 | Load balancing routing method for tree data center link layer based on annealing method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108260169B (en) | QoS guarantee-based dynamic service function chain deployment method | |
| CN107204874A (en) | Ensure the minimum SDN multi-controller dispositions method of time delay | |
| CN104993941B (en) | One kind is based on Openflow network high fault tolerance virtual network mapping algorithms | |
| US20160087848A1 (en) | System, method and apparatus for improving the performance of collective operations in high performance computing | |
| Chen et al. | Multi-controller placement towards SDN based on Louvain heuristic algorithm | |
| JP5943431B2 (en) | Network, data transfer node, communication method and program | |
| CN109286528B (en) | SDN network multi-controller deployment method based on time delay | |
| WO2012149789A1 (en) | Method and device for resource matching in vpc migration | |
| CN104935476B (en) | A kind of network traffics matrix measuring method based on SDN | |
| CN112350949B (en) | Rerouting congestion control method and system based on flow scheduling in software defined network | |
| CN110213363A (en) | Cloud resource dynamic allocation system and method based on software defined network | |
| EP3559833B1 (en) | Best-efforts database functions | |
| CN107276916B (en) | Switch flow table management method based on protocol non-perception forwarding technology | |
| Li et al. | GBC3: A versatile cube-based server-centric network for data centers | |
| CN107800570A (en) | SDN controller dispositions methods based on bat algorithm | |
| CN105308920A (en) | Multicast flow reordering scheme | |
| CN107465966A (en) | A kind of topology reconstruction control method for optical-fiber network | |
| Zhang et al. | NNIRSS: neural network-based intelligent routing scheme for SDN | |
| CN107241277A (en) | Node method for annealing in SDN | |
| US9258254B2 (en) | Virtual router and switch | |
| CN113938434A (en) | Large-scale high-performance RoCEv2 network construction method and system | |
| Guo | Aggregating uncertain incast transfers in BCube-like data centers | |
| Li et al. | Multi-threshold SDN controllers load balancing algorithm based on controller load | |
| Wang et al. | A load-balanced algorithm for multi-controller placement in software-defined network | |
| Dafda et al. | Dynamic load balancing in SDN using energy aware routing and optimization algorithm |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171010 |