CN106656847B - The maximized SDN load-balancing method of network utility - Google Patents
The maximized SDN load-balancing method of network utility Download PDFInfo
- Publication number
- CN106656847B CN106656847B CN201710140102.7A CN201710140102A CN106656847B CN 106656847 B CN106656847 B CN 106656847B CN 201710140102 A CN201710140102 A CN 201710140102A CN 106656847 B CN106656847 B CN 106656847B
- Authority
- CN
- China
- Prior art keywords
- network
- load
- value
- utility
- link
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
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
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
- H04L45/123—Evaluation of link metrics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/302—Route determination based on requested QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/74—Address processing for routing
- H04L45/745—Address table lookup; Address filtering
-
- 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
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
Abstract
The present invention relates to software defined network field, in particular to a kind of maximized software defined network load-balancing method of network utility;Determine mulitpath existing between source node to destination node, according to the difference of data streaming service attribute, assign the different weights for meeting the impact factor of service quality, then pass through the information of acquisition interchanger, the real time information such as the value, including link utilization, time delay and packet loss of impact factor of service quality in network are obtained, the value of utility of each path and load in path set are then calculated, global consideration is finally combined, selects optimal path for data flow;The present invention can meet to the full extent the quality of service requirement of business while guaranteeing network dynamic load balancing.
Description
Technical field
The present invention relates to software defined network (Software Defined Network, SDN), in particular to network utility
Maximized SDN load-balancing method.
Background technique
Software defined network (Software Defined Network, SDN) is Stanford Univ USA Clean Slate
A part in the works is a kind of revolution formula new network framework, is initially campus network researcher's design innovation network rack
Structure simultaneously provides true experiment porch, gradually universal in academia and industrial circle by the popularization of the researchers such as McKeown.?
In SDN network, SDN by network data plane and controls planar separation using the thought of layering, by the centralized control of software-driven
Device carries out control and management in logic to entire SDN network, is responsible to define forward-path, mentions for upper network layer application and service
For programmable interface, there is global network view, can neatly use Internet resources.And data plane only needs to be put down according to control
The forward rule that face generates carries out simple forwarding data flow, reduces the complexity of bottom transmission device function.OpenFlow
Agreement communicates between plane and data plane for controlling, and changes the mode that data packet forwards in traditional network.Data plane
Stream request data package is sent to Centralized Controller by OpenFlow agreement, controller utilizes global network view to data plane
Issue forward rule.Compared with traditional network architecture, this control of SDN forwards the isolated network architecture to pass through Centralized Controller,
Global network state can be monitored in real time, be highly suitable for solving the network resource utilization that current internet faces it is low,
The problems such as link load is unbalanced and network service quality guarantees.SDN is at present in campus network, data center network field
Many achievements are achieved, the Plug-n-Serve Load Balancing Model and Google's number affixed one's name to such as Stamford university computer pastern
According to center B4 network.
Load-balancing technique is balanced by implementing in a network, can effectively improve the throughput and data processing energy of network
Power enhances its availability.Load balancing is a kind of technology that the load by computer network is shared, load
Balanced main purpose is that a large amount of requests for sending outside are evenly dispersed into each section of network according to certain algorithm
In point, so that the computing resource in existing link and network is fully used.It thus can be shortened the response time simultaneously
Improve the throughput of whole network.
Although conventional internet can utilize integrated service (Integrated Services, abbreviation IntServ) or differentiation
It services (Differentiated Service, abbreviation DiffServ) and a degree of service quality (Quality of is provided
Service, abbreviation QoS) it ensures, but since they are built upon on distributed interconnection framework, collect global network state
Information difficulty is huge, therefore can not be promoted in the world.SDN framework will be by that will control plane and data plane point
From, periodically to interchanger send status request information, newest network state information can be obtained.Therefore, SDN being capable of root
Flow is distributed according to real-time link load situation, improves network resource utilization, mitigates network congestion when overload, is guaranteed
QoS.At present according to the distribution object of scheduling, it (is more than certain that the load-balancing technique based on SDN, which can be mainly divided into based on big stream,
Size and long-term stream) identification dispatching algorithm, the scheduling towards QoS and other algorithms.Based on big stream identification
Dispatching algorithm by the big stream of real-time detection network, when congestion occurs for network link, system will be preferably big stream calculation one conjunction
The forward-path of reason guarantees balancing link load to solve link congestion.Flow scheduling towards QoS is then to guarantee to take
Network Load Balance is realized while quality of being engaged in, and mainly for the real-time different business stream in network, calculates a satisfaction for it
The path of QoS demand, when congestion occurs for network, controller will be again in conjunction with service traffics demand and real-time network state
The suitable subsequent path of service selection, to guarantee the QoS demand of business.
Above-mentioned two classes method is all to be calculated in controller Floodlight using Di Jiesitela Dijkstra shortest path
Method is the services selection one shortest path progress data forwarding from source node to destination node, using traditional " minimum hop count "
As link weight, in whole network operational process, the weighted value of link will be invariable, cannot effectively reflect link
Real-time status and quality, are unable to satisfy the qos requirement of user, i.e. the QoS such as network throughput, end-to-end time delay and packet loss influence
The factor cannot reach ideal requirement, and cannot adjust network flow distribution according to the actual loading dynamic of link, cannot achieve
Dynamic load leveling.
Summary of the invention
In order to solve the above technical problems, the invention proposes a kind of maximized software defined network load of network utility is equal
Weighing apparatus method, comprehensively considers QoS of survice feature and link load situation is that data flow carries out Path selection in real time, is guaranteeing that network is negative
Guarantee maximized QoS of survice while carrying balanced.
A kind of maximized software defined network load-balancing method of network utility of the present invention, according to the source IP of data packet
Location and purpose IP address find the set of paths between source node s to destination node d;If set of paths is not sky, number is distinguished
According to the type of service of stream, the given impact factor weight for meeting QoS of survice calculates the influence factor value for meeting QoS of survice, according to
Impact factor weight and influence factor value calculate the value of utility that path meets QoS of survice, calculate the load in path;When network is not gathered around
When plug, select the path that can bring maximum utility for business, when the network is congested, select value of utility and the maximum road of load difference
Diameter.
Preferably, described to be found between source node s to destination node d according to the source IP address and purpose IP address of data packet
Set of paths use k shortest path first.
Preferably, the type of service for distinguishing data flow uses the preceding 6bits in the domain Tos of data packet header.
Preferably, described that the value of utility packet that path meets QoS of survice is calculated according to impact factor weight and influence factor value
It includes:
Indicate the value of utility of path r;
Uij=∑ wnun(xn) indicate the value of utility of link (i, j), wherein n ∈ (u, d, p), wu、wd、wpRespectively bandwidth benefit
With rate loadijValue of utility weighted value, chain-circuit time delay delayijValue of utility weighted value and link packet drop rate lossij's
The weighted value of value of utility;Indicate bandwidth availability ratio loadijValue of utility,Indicate chain-circuit time delay
delayijValue of utility,Indicate link packet drop rate lossijValue of utility;
Indicate bandwidth availability ratio loadijNormalized value,Indicate chain-circuit time delay delayijReturn
One change value,Indicate link packet drop rate lossijNormalized value;Indicate average bandwidth benefit
With rate,Indicate the average delay of path r,Indicate that network link is flat
Equal packet loss;E is the set of link in network (i, j), and m is all chain travel permit numbers.
Preferably, the link packet drop rate calculation includes each measurement period statistics link two ends exchange machine respective end
The statistical data of mouth, with the difference for sending byte number and received byte number of the two port divided by total transmission byte number.
Preferably, the chain-circuit time delay calculation includes:
treceiveIndicate time when controller receives the PACKET_IN message of interchanger j transmission, tsentIndicate controller
Send time of the probe messages to interchanger i, RTTiIt is round-trip delay of the controller to interchanger i, RTTjThen indicate that controller arrives
The round-trip delay of interchanger j.
Preferably, the load for calculating path, including according to link bandwidth utilization rate loadijThe load of path r is calculated,
Include:
Load (r)=Max (loadij)
MaxBandijFor the maximum bandwidth of link (i, j), BandWidtht-interfaceIndicate t moment acquisition interface
The bandwidth that interface inflow direction has occupied, packet_size are acquisition interface interface data package size, Δ t
For flow collection gap periods, T is sample duration.
Preferably, according to link bandwidth utilization rate loadijJudge network whether congestion, including if loadij> 0.7, then
Think that congestion occurs for the link between interchanger i and interchanger j, is otherwise not congestion.
The present invention includes following advantages compared with prior art:
The present invention is the different attribute for meeting QoS of survice, assigns the impact factor of QoS different weights, overcomes existing
The problem of weighted value of link is invariable in technology, cannot effectively reflect the real-time status and quality of link, it is effectively full
The qos requirement of sufficient user;The present invention considers net by combining its respective weights to calculate the value of utility of each path in path set
The real-time condition of network load;When network load is preferable, being selected in the maximum path of value of utility is optimal path, is gathered around in network
When plug, selecting effectiveness and the maximum path of path load difference is optimal path.So as to guarantee Network Load Balance
Simultaneously, moreover it is possible to meet the qos requirement of business to the full extent.
Detailed description of the invention
Fig. 1 is the SDN load-balancing method preferred embodiment flow diagram of inventive network maximization of utility;
Fig. 2 is that the impact factor of present invention selection routing acquires figure;
Fig. 3 is the value of utility calculation flow chart for meeting QoS of survice in the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with attached drawing to of the invention real
Apply example further description.
The software defined network load-balancing method of inventive network maximization of utility, preferred embodiment base as shown in Figure 1
In the load-balancing method preferred embodiment algorithm flow schematic diagram of maximization of utility, can be described as: according to the source of data packet
IP address and purpose IP address find the set of paths between source node s to destination node d;If set of paths is not empty, area
The type of service of divided data stream, the given impact factor weight for meeting QoS of survice, calculates the influence factor value for meeting QoS of survice,
The value of utility that path meets QoS of survice is calculated according to impact factor weight and influence factor value, calculates the load in path;Work as network
Not when congestion, the path that can bring maximum utility for business is selected, when the network is congested, selects value of utility and load difference maximum
Path.
The present invention includes following key step: finding source node to the mulitpath (path set between destination node
It closes);Distinguish the type of service of data flow;Calculate the influence factor value for meeting QoS of survice;According to impact factor weight and influence because
Subvalue calculates the value of utility that path meets QoS of survice;Calculate the load in path;Consider global information and selects Optimization route.Under
It is illustrated respectively in face of each key step.
As the achievable mode of the present invention, the searching source node to the mulitpath between destination node is specifically described
It is as follows:
When a new stream reaches OpenFlow interchanger, there is no matched flow entry in flow table, interchanger will pass through
Data packet is transmitted to controller by PACKET_IN message.When controller receives the PACKET_IN message of interchanger, router-level topology
Module can extract corresponding header packet information, such as source address, destination address, port information to calculate corresponding path.According to mentioning
The IP address of the source node and destination node got is calculated that may be present a plurality of between two nodes using k shortest path first
Routing.
K shortest path first of the present invention stores path candidate using most rickle, by the simplification for pre-processing each link
Expense is spent and is ranked up according to size from being simplified for each node link with corresponding adjacent chained list storage, and extension is candidate every time
The simplification expense the smallest link from deviation point is only chosen when path, and forms kth short path with deflection path.
As the achievable mode of the present invention, the type of service for distinguishing data flow is described in detail below:
Define packet header domain of ten tuples as flow table in OpenFlow standard agreement, wherein IP data packet header
Preceding 6bits is the unitary in ten tuples in the domain Tos, and the service attribute of data flow can be distinguished using the domain Tos.Tos(Type of
Service, type of service) domain refers to that upper layer application is one of IP packet header to the desired service quality of data packet
Point.
Different network services have different needs QoS, it is therefore desirable to consider to reach the business category of data flow in network
Property.The service attribute of data flow can be distinguished according to the domain Tos of IP data packet header, and meets its QoS's according to influence business
The significance level of impact factor assigns corresponding weight.Network service can be divided into session service, streaming media service, interaction
Class business, background business etc..
As the achievable mode of the present invention, the given impact factor weight for meeting QoS of survice is described in detail below:
Requirement of the different business to QoS impact factor is different:
Requirement to time delay is from high to low: session service > streaming media service > interaction service > background business;
Requirement to packet loss is from high to low: background business > interaction service > streaming media service > session service.
It is described to calculate the influence factor value for meeting QoS of survice as the achievable mode of the present invention, it specifically includes:
The impact factor of QoS of the present invention includes the QoS relevant parameter such as Link broadband utilization rate, packet loss and time delay.According to
These parameters, routing calculation module could calculate one according to utility function for business and meet business and network utility maximization
Path, and by obtain link bandwidth utilization rate can monitor network link congestion condition with realize dynamic flow dispatch,
Alleviate network congestion.
Firstly, obtaining the real time data of each interchanger in network by controller, and chain is calculated by collected data
Bandwidth availability ratio, time delay and packet loss between road etc..
Then, the processing these data being normalized, and the Different Effects factor for combining different business to meet its QoS
Respective weights, calculate the value of utility in each path.Meanwhile calculating average load degree and the load of each path of network.
Although OpenFlow agreement provides some functions and reads network state, but these QoS such as time delay, packet loss influence
The factor can not directly acquire, so specific network state information sampling instrument and appraisal procedure is needed to obtain QoS relative influence
The factor.
In order to obtain the real-time and accuracy that guarantee global network filling information, present invention preferably employs active information
Acquisition method periodically initiates each switch port information of inquiry request acquisition of information to interchanger from controller, thus root
The QoS relative influence factor is calculated according to the port information.
1) about the calculating of link utilization
The present invention monitors the flow of OpenFlow interchanger in network using sFlow technology, and is believed according to the flow of statistics
Breath calculates link bandwidth utilization rate, so that controller be avoided frequently to send STATS_REQUEST information to interchanger, reduces network
Expense.The network level real-time traffic monitoring technology based on " statistic sampling mode " that sFlow technology proposes, can not only provide
The whole second layer allows users to grasp network transmission stream in real time to the real-time traffic information in the 4th layer even network-wide basis
Performance with and situation.SFlow monitoring tools are made of sFlow Agent and sFlowCollector two parts.
SFlow Agent is mounted on OpenFLow interchanger by load-balancing mechanism proposed by the present invention
In, it obtains the packet information for entering exchange interface and is packaged into sFlow message, be sent to and be integrated in Floodlight
sFlow Collector.SFlow Collector can receive the sFlow message that multiple sFlow Agent are sent simultaneously, and
To the processing for statistical analysis of sFlow message.Then the exchange interface number that controller is then counted according to sFlow Collector
According to calculating the bandwidth of each exchange interface by following formula.
In formula, BandWidtht-interfaceIndicate the band that t moment acquisition interface interface inflow direction has occupied
Width, packet_size are sFlow Agent acquisition interface interface data package size, and Δ t is flow collection gap periods,
T is sample duration.
Therefore, for the bandwidth availability ratio load of the link (i, j) between interchanger i and interchanger jijIt calculates such as following formula institute
Show:
In formula, MaxBandijFor the maximum bandwidth of link (i, j).Therefore, network state acquisition module then can be obtained periodically
Take the service condition of each of the links in network.
Congestion monitoring modular determines whether link occurs congestion by checking link utilization, if load(i,j)> 0.7, then
Think that congestion occurs for the link between interchanger i and interchanger j, congestion monitoring modular will send congestion signal and give router-level topology mould
Block.
2) about the calculating of time delay
Time delay is the very important service impact factor in QoS, especially to the very high multimedia service of requirement of real-time,
Such as VoIP and interactivity video traffic, it is desirable that time delay is no more than 150ms end to end, once it is more than the service experience of user
By degradation.
Network parameter measurement method that the present invention is provided using OpenNetMon calculates chain-circuit time delay and packet loss.Such as Fig. 3
It is shown, specifically include following procedure:
Controller periodically injects link layer discovery association to interchanger i by PACKET_OUT message in OpenNetMon scheme
LLDP broadcast packet is discussed, sending time t is recordedsend, LLDP broadcast packet will in interchanger i install a new flow table, show by
Packet is sent to interchanger j.
When interchanger j receives the LLDP packet, due to not having matching flow table in interchanger j, old friend changes planes j will be to
Controller sends PACKET_IN message.
After controller receives the PACKET_IN message of interchanger j transmission, receiving time t is recordedrecieve, then can obtain
Probe messages are sent to the total time for receiving PACKET_IN message from controller out.Due to controller always with the friendship in network
It changes planes the connection kept constant, controller can recorde corresponding to receiving to interchanger from FEATURE_REQUEST message is sent
Round-trip delay when FEATURE_REPLY message, to know oneself time delay between interchanger.
So between interchanger i and interchanger j link (i, j) time delay delayijIt is obtained by following formula:
In formula, treceiveIndicate time when controller receives the PACKET_IN message of interchanger j transmission, tsentIndicate control
Device processed sends time of the probe messages to interchanger i, RTTiIt is round-trip delay of the controller to interchanger i, RTTjIt then indicates to control
Round-trip delay of the device to interchanger j.Wherein, round-trip delay RTT of the controller to interchangerswitchIt is calculate by the following formula to obtain:
RTTswitch=tr-ts
Wherein, tsIndicate that controller sends STATS_REQUEST message to the time of interchanger, and trThen indicate controller
Receive the time of STATS_REPLY message.
3) about calculating link packet drop rate
The interchanger of a both link ends is calculated not instead of as unit of stream to the statistics of packet loss in OpenNetMon
Statistical data, controller periodically distinguishes the interchanger at interrogation link both ends, obtains the connected port meter of two interchangers
The statistical data of number device.Prior art OpenNetMon is to count link packet drop rate to each period, i.e., in measurement period
The absolute value that the increment of two ends exchange machine counter subtracts each other is exactly the packet loss of this link divided by measurement period.
Particularly, the present invention improves OpenNetMon packet loss measurement method, and each measurement period counts link two
The statistical data of end switch corresponding port, with the difference for sending byte number and received byte number of the two port divided by total hair
Sending byte number is exactly the packet loss of this link, and calculation formula is as follows:
In formula, receivei、receivejInterchanger i is respectively indicated, j is in link lijThe byte number of upper transmission, sendi、
sendjRespectively indicate interchanger i, j received byte number on link (i, j).
The present invention is about the value of utility and network real time load for calculating path, in the following ways.
Network representation is a weighted graph G (N, E) by the present invention first, such as shown in Fig. 2, wherein N={ node1,
node2,...,nodenIndicate network in interchanger set, E be Fig. 2 in side set, indicate interchanger between link
(set of link in network), and it is n that node number, which is arranged, number of edges is m (quantity of link).Therefore side eijIndicate interchanger i
With the link between interchanger j.According to the network state information of acquisition, each edge e in G (N, E) is schemedijWith a 3 dimensional vector tables
Show its link state, i.e. xij=(delayij,lossij,loadij), wherein delayij,lossij,loadijRespectively indicate exchange
Time delay, packet loss and link utilization between machine i and interchanger j.It will be from certain source node s to destination node d
Path is denoted as r, and r ∈ E, then the formalization representation of the impact factor of QoS is as follows:
(1) time delay of path r
(2) average delay of path r
(3) packet loss of path r
(4) network link average packet loss ratio
(5) averaging bandwidth utilization
(6) link load is shaken
In order to meet meet business data transmission while guaranteeing Network Load Balance needed for qos requirement, need to examine
The relative influences factors such as time delay, packet loss and the utilization rate of link are considered, with the forward-path for selecting current network environment optimal.From
From the point of view of data forwarding plane, the time delay of conversion link reflects the efficiency of transmission of current ink, and the packet loss of conversion link is then
The reliability for reflecting link data transfer should select that time delay is small, packet loss according to the different QoS requirements of different business as far as possible
Lower link, to ensure QoS.Meanwhile in order to guarantee network traffic load equilibrium, business should select available bandwidth compared with
Big link carries out data forwarding.Therefore, the load-balancing method proposed by the present invention based on maximization of utility considers not of the same trade or business
The network utility value for each path that can be obtained from set of paths of being engaged in.Network effect based on different business attribute QoS perception
The link weight of link-quality superiority and inferiority between interchanger can be described with value U (x), value is bigger, indicates that link can be maximumlly
The QoS demand of guarantee business.
For new data stream, utility function is defined as U (x), wherein x represent influence Business Stream QoS it is each because
The vector of element composition, the satisfaction that Business Stream is obtained from network is measured by the value of U (x).Effectiveness computation model such as Fig. 3 institute
Show.About multiattribute utility function selection its be defined as follows shown in formula:
In formula, x is the vector of n attribute composition, ui(xi) it is attribute xiCorresponding value of utility, wiFor corresponding weight, power
Weight can be adjusted according to the qos requirement of practical business dynamic, since link in network packet loss is between 0 to 1, and chain-circuit time delay with
Link utilization, which is compared, belongs to big number, widely different on the order of magnitude, therefore needs first to carry out when carrying out link weight calculating
Normalized.The present invention is using to link bandwidth utilization rate, time delay, packet loss normalized, calculating is such as following formula institute respectively
Show:
uij, dij, pijIndicate link bandwidth utilization rate loadij, chain-circuit time delay delayijWith link packet drop rate lossijIt carries out
Numerical value after normalized.
For Business Stream, time delay is smaller on the link for it, and packet loss is smaller, and bandwidth availability ratio is smaller, meets the QoS of business
Value of utility is better, therefore bandwidth availability ratio loadij, chain-circuit time delay delayijWith link packet drop rate lossijValue of utility difference
Are as follows:
It is experienced for different business stream and the different QoS of requirement, the present invention can dynamically change the weight of three, with full
Sufficient business obtains maximum effectiveness.So the value of utility U of link (i, j)ijFor
Uij=∑ wnun(xn) wherein n ∈ (u, d, p)
Wherein wu、wd、wpRespectively bandwidth availability ratio loadij, chain-circuit time delay delayijWith link packet drop rate lossijEffect
With the corresponding weighted value of value.
Therefore, the present invention, which is defined on obtainable total utility U (r) on the r of path and can obtain for user on path, meets business
The effectiveness of QoS is indicated by the sum of QoS effectiveness of every section of link, is shown below:
The link bandwidth utilization rate load arrived calculated according to frontij, the load of path r can be calculated, is shown below:
Load (r)=Max (loadij),(i,j)∈r
For the present invention for judging whether network occurs congestion, and using the value of utility in the path and load, selection is best
Path, it is preferred to use following manner:
According to the link bandwidth utilization rate load being the previously calculatedij, it is possible to determine that whether link occurs congestion, if
loadij> 0.7, then it is assumed that congestion occurs for the link between interchanger i and interchanger j, is otherwise not congestion.
Finally consider the above global information, select optimal path: judging network not congestion, selects value of utility in path set
Maximum path.Judge that network generates congestion, selects value of utility and the maximum path of load difference in path set.
In network not congestion, the maximum path of effectiveness in path set is selected, can satisfy the maximum QoS of user in this way.
Congestion is generated in network, the reason of value of utility maximum with load difference path is that load is according to link bandwidth in choice set
What utilization rate was got, if link bandwidth utilization rate is high, its reciprocal value is taken when calculating effectiveness, its effectiveness will become
It is small.Therefore, then big load is subtracted, the difference in path can become smaller certainly.On the contrary, difference turn out greatly load it is small.That
The selection maximum path of difference is equal to select not only to meet a load balancing, but also meets the maximized path QoS.
The present invention is when the load criteria variance of network is not above threshold value, i.e., when equilibrium degree is preferable, selects to be business
Bring the path of maximum utility max (U (r)).When network equilibrium degree is poor, i.e., when network load is unbalanced, the present invention selects effect
With with the maximum path of load difference.In this way, meeting the QoS of business when network load is more balanced, being provided for network
Best experience.When congestion occurs for network, while QoS of survice and network load are considered, is guaranteeing Network Load Balance
Meanwhile obtaining maximized QoS of survice.
The present invention determines mulitpath existing between source node to destination node first, according to data streaming service attribute
Difference, assign meet service quality impact factor different weights, then by acquisition interchanger information, obtain net
Influence factor value of service quality in network, including link utilization, time delay and packet loss etc. then calculate in path set every
The value of utility in path and load, finally combine global consideration, select optimal path for data flow.The present invention can guarantee network
While dynamic load leveling, meet the service quality of business to the full extent.
Embodiment provided above has carried out further detailed description, institute to the object, technical solutions and advantages of the present invention
It should be understood that embodiment provided above is only the preferred embodiment of the present invention, be not intended to limit the invention, it is all
Any modification, equivalent substitution, improvement and etc. made for the present invention, should be included in the present invention within the spirit and principles in the present invention
Protection scope within.
Claims (8)
1. a kind of maximized software defined network load-balancing method of network utility, it is characterised in that: according to the source of data packet
IP address and purpose IP address find source node to the set of paths between destination node;If set of paths is not sky, distinguish
The type of service of data flow, the given impact factor weight for meeting QoS of survice, calculates the influence factor value for meeting QoS of survice, root
The value of utility that path meets QoS of survice is calculated according to impact factor weight and influence factor value, calculates the load in path;When network not
When congestion, select the path that can bring maximum utility for business, when the network is congested, select value of utility and load difference maximum
Path;
The impact factor of QoS includes Link broadband utilization rate, packet loss and time delay;
It is described path calculated according to impact factor weight and influence factor value meet the value of utility of QoS of survice include:
Indicate the value of utility of path r;
Uij=∑ wnun(xn) indicate the value of utility of link (i, j), wherein n ∈ (u, d, p), wu、wd、wpRespectively bandwidth availability ratio
loadijValue of utility weighted value, chain-circuit time delay delayijValue of utility weighted value and link packet drop rate lossijEffectiveness
The weighted value of value;Indicate bandwidth availability ratio loadijValue of utility,Indicate chain-circuit time delay
delayijValue of utility,Indicate link packet drop rate lossijValue of utility;
Indicate bandwidth availability ratio loadijNormalized value,Indicate chain-circuit time delay delayijNormalization
Value,Indicate link packet drop rate lossijNormalized value;Indicate that average bandwidth utilizes
Rate,Indicate the average delay of path r,Indicate that network link is average
Packet loss;E is the set of link in network (i, j), and m is all chain travel permit numbers.
2. the maximized software defined network load-balancing method of network utility according to claim 1, it is characterised in that: institute
State according to the source IP address and purpose IP address of data packet find source node to the set of paths between destination node using k it is most short
Routing algorithm.
3. the maximized software defined network load-balancing method of network utility according to claim 1, it is characterised in that: institute
State distinguish data flow type of service using data packet header the domain Tos preceding 6bits.
4. according to claim 1 or the 3 maximized software defined network load-balancing methods of network utility, feature exist
In: the type of service includes session service, streaming media service, interaction service and background business.
5. the maximized software defined network load-balancing method of network utility according to claim 1, it is characterised in that: institute
The statistical data that link packet drop rate calculation includes each measurement period statistics link two ends exchange machine corresponding port is stated, with two
The difference for sending byte number and received byte number of person port is divided by total transmission byte number.
6. the maximized software defined network load-balancing method of network utility according to claim 1, it is characterised in that: institute
Stating chain-circuit time delay calculation includes:
treceiveIndicate time when controller receives the PACKET_IN message of interchanger j transmission, tsentIndicate that controller is sent
Time of the probe messages to interchanger i, RTTiIt is round-trip delay of the controller to interchanger i, RTTjThen indicate controller to exchange
The round-trip delay of machine j.
7. the maximized software defined network load-balancing method of network utility according to claim 1, it is characterised in that: institute
The load for calculating path is stated, including according to link bandwidth utilization rate loadijCalculate the load of path r, comprising:
Load (r)=Max (loadij)
MaxBandijFor the maximum bandwidth of link (i, j), BandWidtht-interfaceIndicate t moment acquisition interface interface
The bandwidth that inflow direction has occupied, packet_size are acquisition interface interface data package size, and Δ t is flow collection
Gap periods, T are sample duration.
8. the maximized software defined network load-balancing method of network utility according to claim 1, it is characterised in that: root
According to link bandwidth utilization rate loadijJudge network whether congestion, including if loadij> 0.7, then it is assumed that interchanger i and exchange
Congestion occurs for the link between machine j, is otherwise not congestion.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710140102.7A CN106656847B (en) | 2017-03-10 | 2017-03-10 | The maximized SDN load-balancing method of network utility |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710140102.7A CN106656847B (en) | 2017-03-10 | 2017-03-10 | The maximized SDN load-balancing method of network utility |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106656847A CN106656847A (en) | 2017-05-10 |
CN106656847B true CN106656847B (en) | 2019-05-24 |
Family
ID=58848156
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710140102.7A Active CN106656847B (en) | 2017-03-10 | 2017-03-10 | The maximized SDN load-balancing method of network utility |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106656847B (en) |
Families Citing this family (63)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10454714B2 (en) | 2013-07-10 | 2019-10-22 | Nicira, Inc. | Method and system of overlay flow control |
US10498652B2 (en) | 2015-04-13 | 2019-12-03 | Nicira, Inc. | Method and system of application-aware routing with crowdsourcing |
US10135789B2 (en) | 2015-04-13 | 2018-11-20 | Nicira, Inc. | Method and system of establishing a virtual private network in a cloud service for branch networking |
US20180219765A1 (en) | 2017-01-31 | 2018-08-02 | Waltz Networks | Method and Apparatus for Network Traffic Control Optimization |
US11706127B2 (en) | 2017-01-31 | 2023-07-18 | Vmware, Inc. | High performance software-defined core network |
US20200036624A1 (en) | 2017-01-31 | 2020-01-30 | The Mode Group | High performance software-defined core network |
US10992568B2 (en) | 2017-01-31 | 2021-04-27 | Vmware, Inc. | High performance software-defined core network |
US10523539B2 (en) | 2017-06-22 | 2019-12-31 | Nicira, Inc. | Method and system of resiliency in cloud-delivered SD-WAN |
CN107294852B (en) * | 2017-07-01 | 2019-12-27 | 台州学院 | Network routing method using topology dispersed short path set |
CN107395697B (en) * | 2017-07-06 | 2021-01-26 | 创新先进技术有限公司 | Push channel selection method, message push method, device and equipment, and readable medium |
CN107483337A (en) * | 2017-07-31 | 2017-12-15 | 郑州云海信息技术有限公司 | Determine the method and its network equipment in path |
CN107370676A (en) * | 2017-08-03 | 2017-11-21 | 中山大学 | Fusion QoS and load balancing demand a kind of route selection method |
CN107508730B (en) * | 2017-09-29 | 2022-02-18 | 刘昱 | SDN network-based data center interconnection method and device |
CN107484209B (en) * | 2017-09-30 | 2020-10-09 | 南京南瑞集团公司 | Network load balancing vertical switching method considering user QoS |
US11115480B2 (en) | 2017-10-02 | 2021-09-07 | Vmware, Inc. | Layer four optimization for a virtual network defined over public cloud |
US11005684B2 (en) | 2017-10-02 | 2021-05-11 | Vmware, Inc. | Creating virtual networks spanning multiple public clouds |
US10999100B2 (en) | 2017-10-02 | 2021-05-04 | Vmware, Inc. | Identifying multiple nodes in a virtual network defined over a set of public clouds to connect to an external SAAS provider |
CN107770065A (en) * | 2017-10-10 | 2018-03-06 | 山东大学 | A kind of streaming media path selecting system based on software defined network SDN |
US11223514B2 (en) | 2017-11-09 | 2022-01-11 | Nicira, Inc. | Method and system of a dynamic high-availability mode based on current wide area network connectivity |
CN107948067B (en) * | 2017-11-13 | 2020-10-20 | 西安邮电大学 | Link load balancing method for QoS guarantee of multiple service flows in software defined network |
CN108199970B (en) * | 2017-12-15 | 2020-08-04 | 浙江大学 | Method for reconstructing data packet path in software defined network |
CN108322406B (en) * | 2017-12-28 | 2021-10-08 | 广东电网有限责任公司电力调度控制中心 | SDN data plane fault recovery method based on link performance and flow classification |
CN110011915A (en) * | 2018-01-04 | 2019-07-12 | 杭州达乎科技有限公司 | SDN network framework, flow transmission control method and device based on SDN |
CN108337184A (en) * | 2018-01-11 | 2018-07-27 | 山石网科通信技术有限公司 | Flow equalization dispatching method and device |
CN110113222B (en) * | 2018-02-01 | 2023-05-02 | 中兴通讯股份有限公司 | Method and device for acquiring link bandwidth utilization rate and terminal |
WO2019241992A1 (en) * | 2018-06-22 | 2019-12-26 | 华为技术有限公司 | Routing method, node and computer storage medium |
CN110838979B (en) * | 2018-08-17 | 2022-02-08 | 中国电信股份有限公司 | Flow forwarding control method, device, system and computer readable storage medium |
CN109600322A (en) * | 2019-01-11 | 2019-04-09 | 国家电网有限公司 | A kind of flow load balance control method suitable for electric power backbone communications |
CN111756557B (en) * | 2019-03-26 | 2023-05-23 | 华为技术有限公司 | Data transmission method and device |
CN110365582A (en) * | 2019-06-20 | 2019-10-22 | 山东省计算中心(国家超级计算济南中心) | A kind of multiple constraint method for routing based on SDN network, a kind of controller |
CN110311864A (en) * | 2019-06-20 | 2019-10-08 | 山东省计算中心(国家超级计算济南中心) | Method for routing and device based on entropy assessment in a kind of SDN network |
CN110412983B (en) * | 2019-08-01 | 2022-11-01 | 北京百度网讯科技有限公司 | Vehicle collision avoidance detection method and device and vehicle |
US11153230B2 (en) | 2019-08-27 | 2021-10-19 | Vmware, Inc. | Having a remote device use a shared virtual network to access a dedicated virtual network defined over public clouds |
CN110535770B (en) * | 2019-08-30 | 2021-10-22 | 西安邮电大学 | QoS-aware-based intelligent routing method for video stream in SDN environment |
CN110752993B (en) * | 2019-10-24 | 2022-02-25 | 新华三信息安全技术有限公司 | Message forwarding method and device |
US11611507B2 (en) | 2019-10-28 | 2023-03-21 | Vmware, Inc. | Managing forwarding elements at edge nodes connected to a virtual network |
CN111082978A (en) * | 2019-12-05 | 2020-04-28 | 西安电子科技大学 | SDN-oriented utility bandwidth allocation method |
US11489783B2 (en) | 2019-12-12 | 2022-11-01 | Vmware, Inc. | Performing deep packet inspection in a software defined wide area network |
US11394640B2 (en) | 2019-12-12 | 2022-07-19 | Vmware, Inc. | Collecting and analyzing data regarding flows associated with DPI parameters |
CN110995593A (en) * | 2019-12-18 | 2020-04-10 | 深信服科技股份有限公司 | Intelligent information forwarding method, device, equipment and storage medium |
US11606712B2 (en) | 2020-01-24 | 2023-03-14 | Vmware, Inc. | Dynamically assigning service classes for a QOS aware network link |
CN111343097B (en) * | 2020-02-29 | 2023-04-18 | 杭州迪普科技股份有限公司 | Link load balancing method and device, electronic equipment and storage medium |
CN111447152B (en) * | 2020-03-16 | 2023-10-13 | Oppo广东移动通信有限公司 | Sub-stream resource scheduling method, device, terminal equipment and storage medium |
US11245641B2 (en) | 2020-07-02 | 2022-02-08 | Vmware, Inc. | Methods and apparatus for application aware hub clustering techniques for a hyper scale SD-WAN |
US11363124B2 (en) | 2020-07-30 | 2022-06-14 | Vmware, Inc. | Zero copy socket splicing |
CN113286269B (en) * | 2020-10-31 | 2023-03-07 | 国网河南省电力公司经济技术研究院 | Electric power communication network resource management method based on edge calculation |
US11444865B2 (en) | 2020-11-17 | 2022-09-13 | Vmware, Inc. | Autonomous distributed forwarding plane traceability based anomaly detection in application traffic for hyper-scale SD-WAN |
US11575600B2 (en) | 2020-11-24 | 2023-02-07 | Vmware, Inc. | Tunnel-less SD-WAN |
US11601356B2 (en) | 2020-12-29 | 2023-03-07 | Vmware, Inc. | Emulating packet flows to assess network links for SD-WAN |
WO2022154850A1 (en) * | 2021-01-18 | 2022-07-21 | Vmware, Inc. | Network-aware load balancing |
CN116783874A (en) | 2021-01-18 | 2023-09-19 | Vm维尔股份有限公司 | Network aware load balancing |
CN114827018B (en) * | 2021-01-28 | 2023-09-08 | 北京邮电大学 | SDN route selection method and device based on service class and electronic equipment |
CN113067863B (en) * | 2021-03-17 | 2022-04-19 | 西安电子科技大学 | Dynamic load balancing method based on software defined network |
US11381499B1 (en) | 2021-05-03 | 2022-07-05 | Vmware, Inc. | Routing meshes for facilitating routing through an SD-WAN |
US11729065B2 (en) | 2021-05-06 | 2023-08-15 | Vmware, Inc. | Methods for application defined virtual network service among multiple transport in SD-WAN |
CN113114582B (en) * | 2021-05-25 | 2022-05-17 | 电子科技大学 | Link congestion fault prediction and network autonomous control method based on machine learning |
CN113542051B (en) * | 2021-05-27 | 2022-08-30 | 贵州电网有限责任公司 | Monitoring flow load balancing method based on software defined network |
US11489720B1 (en) | 2021-06-18 | 2022-11-01 | Vmware, Inc. | Method and apparatus to evaluate resource elements and public clouds for deploying tenant deployable elements based on harvested performance metrics |
US11943146B2 (en) | 2021-10-01 | 2024-03-26 | VMware LLC | Traffic prioritization in SD-WAN |
CN113766455B (en) * | 2021-11-09 | 2022-04-01 | 迪泰(浙江)通信技术有限公司 | Shipborne communication network switching control method and controller |
CN114007250B (en) * | 2021-12-30 | 2022-06-24 | 中国铁道科学研究院集团有限公司通信信号研究所 | Multi-mobile-network access selection method for railway application |
US11909815B2 (en) | 2022-06-06 | 2024-02-20 | VMware LLC | Routing based on geolocation costs |
CN116647878A (en) * | 2023-07-21 | 2023-08-25 | 北京最终前沿深空科技有限公司 | Load balancing method, device, equipment and storage medium for satellite network |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102055663A (en) * | 2009-11-10 | 2011-05-11 | 武汉大学 | QoS (Quality of Service) route distributing method for realizing load balance in overlay network |
CN102630093A (en) * | 2012-05-02 | 2012-08-08 | 北京邮电大学 | Network access method for load balance in isomer wireless network |
CN105721302A (en) * | 2016-03-31 | 2016-06-29 | 重庆邮电大学 | Routing selection method for software defined network |
-
2017
- 2017-03-10 CN CN201710140102.7A patent/CN106656847B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102055663A (en) * | 2009-11-10 | 2011-05-11 | 武汉大学 | QoS (Quality of Service) route distributing method for realizing load balance in overlay network |
CN102630093A (en) * | 2012-05-02 | 2012-08-08 | 北京邮电大学 | Network access method for load balance in isomer wireless network |
CN105721302A (en) * | 2016-03-31 | 2016-06-29 | 重庆邮电大学 | Routing selection method for software defined network |
Non-Patent Citations (1)
Title |
---|
Combining OpenFlow and sFlow for an effective and scalable anomaly detection and mitigation mechanism on SDN environments;Giotis K, Argyropoulos C, Androulidakis G, et al;《Computer Networks the International Journal of Computer &》;20141231;全文 |
Also Published As
Publication number | Publication date |
---|---|
CN106656847A (en) | 2017-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106656847B (en) | The maximized SDN load-balancing method of network utility | |
CN108600102B (en) | Flexible data transmission system based on intelligent cooperative network | |
CN106411733B (en) | A kind of SDN dynamic load leveling dispatching method based on link real time load | |
WO2017005223A1 (en) | Information sharing method of smart electricity meter, smart electricity meter and acquisition router | |
CN106059821B (en) | A kind of data center traffic QoS guarantee method based on SDN | |
CN107370676A (en) | Fusion QoS and load balancing demand a kind of route selection method | |
Su et al. | CeMon: A cost-effective flow monitoring system in software defined networks | |
Liu et al. | Delay-optimized video traffic routing in software-defined interdatacenter networks | |
CN109756421B (en) | Congestion control system and method based on OpenFlow technology | |
Trestian et al. | OFLoad: An OpenFlow-based dynamic load balancing strategy for datacenter networks | |
CN109547341B (en) | Load sharing method and system for link aggregation | |
US11184284B2 (en) | Data packet forwarding method and apparatus | |
CN103416022B (en) | Throughput testing approach and system in service in distribution router/exchange board structure | |
CN105791151B (en) | A kind of dynamic flow control method and device | |
CN106550049B (en) | A kind of Middleware portion arranging method, apparatus and system | |
CN112491619A (en) | Self-adaptive distribution technology for service customized network resources based on SDN | |
CN101815002B (en) | Method for evaluating overall performance of network based on multi-measure deviation maximization | |
CN109005126B (en) | Data stream processing method, device and computer readable storage medium | |
Zaw et al. | Traffic management with elephant flow detection in software defined networks (SDN) | |
Xu et al. | Partial flow statistics collection for load-balanced routing in software defined networks | |
Wang et al. | An ACO-based elephant and mice flow scheduling system in SDN | |
CN103414594A (en) | IP stream information statistical method for charging and monitoring | |
US10523571B2 (en) | Processing data items in a communications network | |
CN105262682B (en) | A kind of software defined network system and its traffic grooming method for electric power data communication | |
JP2008048131A (en) | P2p traffic monitoring and control system, and method therefor |
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 |