CN106375131B - A kind of virtualization network uplink load-balancing method - Google Patents
A kind of virtualization network uplink load-balancing method Download PDFInfo
- Publication number
- CN106375131B CN106375131B CN201610913667.XA CN201610913667A CN106375131B CN 106375131 B CN106375131 B CN 106375131B CN 201610913667 A CN201610913667 A CN 201610913667A CN 106375131 B CN106375131 B CN 106375131B
- Authority
- CN
- China
- Prior art keywords
- load
- physical network
- message
- network card
- uplink
- 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
- 238000000034 method Methods 0.000 title claims abstract description 17
- 230000005540 biological transmission Effects 0.000 claims abstract description 10
- 238000010276 construction Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000011144 upstream manufacturing Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000005303 weighing Methods 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
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
- H04L41/08—Configuration management of networks or network elements
- H04L41/0896—Bandwidth or capacity management, i.e. automatically increasing or decreasing capacities
Abstract
The present invention is more particularly directed to a kind of virtualization network uplink load-balancing methods.The virtualization network uplink load-balancing method, structure include load counter, load balancer and load container, and the load counter and load container are all connected to load balancer, and the load balancer is connected to each physical network card;The load balancer is sent next message by calculating selection physical network card using the message number aggregate-value and physical network card number of load counter and load container acquisition, and calls the transmission for sending function and realizing message of each physical network card.The virtualization network uplink load-balancing method, using simplest logical algorithm, make adjacent message pass through different physical network cards respectively to issue, it avoids occupying a large amount of cpu resources, method is succinct efficiently easily to be realized, load balancing can be reached between the physical network card that uplink is bound, use physical network card upstream bandwidth to greatest extent.
Description
Technical field
The present invention relates to virtualization network design technical field, in particular to a kind of virtualization network uplink load is equal
Weighing apparatus method.
Background technique
In internet, the world today of cloud computing, big data rapid development, the network throughput of server is also at full speed therewith
Increase, can quickly and the more services of offer are it is necessary to carry more flows.But if in order to improve some handling capacities,
And more advanced network card equipment is purchased, upgraded, server use cost will necessarily improve;Meanwhile with server itself
The network card equipment that performance limitation can be supported also has limitation.
The heart forms a uplink being aggregating usually by the way that muti-piece physical network card to be held together in the server
Road, to achieve the purpose that expand bandwidth.If muti-piece physical network card is bound together the uplink to form a bigger bandwidth,
When there is message to send, it is necessary to which a strategy is allowed to send from different network interface cards, to achieve the effect that load balancing;And
If only redundancy, bandwidth cannot increase.Using the strategy of load balancing, bandwidth can not only be increased, equally had
The effect of redundancy.
Data packet can not be really evenly distributed to each by virtualization Network Load Balance algorithm there are certain defect at present
A physical network card.As famous openvswitch provides three kinds of load-balancing methods: balance-slb, balance-tcp,
Active-backup, active-backup mode can only redundancy, be unable to load balancing;And balance-slb and balance-
Although both modes of tcp can be according to L2, L3 and L4 layers of progress load balancing, both algorithms are fixed Yi Dan calculated
The stream of feature will only be sent from some network interface card, when only one stream or have several stream features it is identical when all will be from certain
One network interface card is sent, and has certain defect.
Based on the above situation, it effectively solves the above problems in order to balancing flow to different network interface cards, the invention proposes
A kind of virtualization network uplink load-balancing method.
Summary of the invention
It is negative that in order to compensate for the shortcomings of the prior art, the present invention provides a kind of virtualization network uplinks being simple and efficient
Carry equalization methods.
The present invention is achieved through the following technical solutions:
A kind of virtualization network uplink load-balancing method, it is characterised in that: building uplink sends message and patrols
Volume structure, it includes load counter, load balancer and load container that the uplink, which sends message logical construction, described negative
It carries counter and load container is all connected to load balancer, the load balancer is connected to each physical network card;It is described negative
It carries counter and adds up before entering load balancer to each message, when the physical network card composition of uplink changes
When load counter is zeroed out;The load container records uplink physical network interface card number, when uplink physical net
When card composition changes, real-time change is carried out;The data that the load balancer is acquired using load counter and load container,
Next message is sent by calculating selection physical network card, and the transmission function of each physical network card is called to realize message
Transmission.
The initial value of the load counter is 0, when message will be sent out message from this uplink, increases by 1;
The initial value of the load container is also 0, and when one physical network card of every increase increases by 1, and when one physical network card of every reduction subtracts 1;
The load balancer sends function to each physical network card and n is successively numbered, and n is natural number, and according to algorithms selection phase
The physical network card answered sends function, message is sent, the corresponding transmission function of the physical network card that number is n is ethn_tx.
When reaching a message, will pass, by load counter count=count+1;Message enters load
?
Weighing apparatus module, load balancer obtain message number aggregate-value count from load counter and load container respectively
And object
Network interface card number m is managed, passes through algorithm: Select=(count%m)+1;
If Selcet is equal to 1, calls eth1_tx to send function and send message;If Select=2 calls eth2_tx
Transmission function send message;If Select is equal to n, calls ethn_tx to send function and send message.eth1_tx,eth2_
Tx, ethn_tx respectively correspond the physical network card that uplink number is 1,2, n.
The beneficial effects of the present invention are: the virtualization network uplink load-balancing method, using simplest logic
Algorithm makes adjacent message pass through different physical network cards respectively and issues, avoids occupying a large amount of cpu resources, method is succinctly efficient
It easily realizes, load balancing can be reached between the physical network card that uplink is bound, use physical network card uplink to greatest extent
Bandwidth.
Detailed description of the invention
Attached drawing 1 is that uplink of the present invention sends message logical construction schematic diagram.
Specific embodiment
In order to which technical problems, technical solutions and advantages to be solved are more clearly understood, tie below
Drawings and examples are closed, the present invention will be described in detail.It should be noted that specific embodiment described herein is only used
To explain the present invention, it is not intended to limit the present invention.
The virtualization network uplink load-balancing method constructs a uplink first and sends message logic knot
Structure, it includes load counter, load balancer and load container, the load meter that the uplink, which sends message logical construction,
Number device and load container are all connected to load balancer, and the load balancer is connected to each physical network card;The load meter
Number devices to each message enter load balancer before add up, when uplink physical network card composition change when pair
Load counter is zeroed out;The load container records uplink physical network interface card number, when uplink physical network interface card group
When at changing, real-time change is carried out;The data that the load balancer is acquired using load counter and load container, pass through
It calculates selection physical network card to send next message, and calls the hair for sending function and realizing message of each physical network card
It send.
The initial value of the load counter is 0, when message will be sent out message from this uplink, increases by 1;
The initial value of the load container is also 0, and when one physical network card of every increase increases by 1, and when one physical network card of every reduction subtracts 1;
The load balancer sends function to each physical network card and is successively numbered 1,2,3 ..., and ..n, n are natural number, and according to
The corresponding physical network card of algorithms selection sends function, message is sent, the corresponding transmission function of the physical network card that number is n
For ethn_tx.
Specific calculation method is as follows:
When reaching a message, will pass, by load counter count=count+1;Message enters load
Equalizer module, load balancer obtain message number aggregate-value count and physics from load counter and load container respectively
Network interface card number m, passes through algorithm: Select=(count%m)+1;
If Selcet is equal to 1, calls eth1_tx to send function and send message;If Select=2 calls eth2_tx
Transmission function send message;If Select is equal to n, calls ethn_tx to send function and send message.eth1_tx,eth2_
Tx, ethn_tx respectively correspond the physical network card that uplink number is 1,2, n.
The virtualization network uplink load-balancing method is suitable for the various band networks such as linux, window, unix
Operating system, the message that server is sent by uplink, successively uplink binding physical network card it
Between carry out rotation transmission, the number for making each network interface card send message reaches balanced to reach flow equalization.Although message has greatly
Have small, but this has no effect on the equilibrium of flow.And since each application program in upper layer is sent out message, when needing to obtain CPU
Between piece, therefore each application program is also to be sent out message by turns, those of is being sent in timeslice, message size is basic
It is consistent.And when flow reaches network interface card maximum ability to take the burden, for a transmission 1G/S flow, message interval is no more than
It 0.0014 millisecond, if there is 2 physical network cards successively to send packet in turn, is equivalent to and entire flow is sampled for 0.0028 second.It is right
In this two groups of samplings that this two physical network cards obtain, mean value error says the power of this flow equalization less than one thousandth
Degree has been able to reach real equilibrium.
Claims (1)
1. a kind of virtualization network uplink load-balancing method, it is characterised in that: building uplink sends message logic
Structure, it includes load counter, load balancer and load container, the load that the uplink, which sends message logical construction,
Counter and load container are all connected to load balancer, and the load balancer is connected to each physical network card;The load
Counter adds up before entering load balancer to each message, when the physical network card of uplink composition changes
Load counter is zeroed out;The load container records uplink physical network interface card number, when uplink physical network interface card
When composition changes, real-time change is carried out;The data that the load balancer is acquired using load counter and load container, lead to
It crosses calculating selection physical network card to send next message, and the function that sends of each physical network card is called to realize message
It sends;
The initial value of the load counter is 0, when message will be sent out message from this uplink, increases by 1;It is described
The initial value of load container is also 0, and when one physical network card of every increase increases by 1, and when one physical network card of every reduction subtracts 1;It is described
Load balancer sends function to each physical network card and n is successively numbered, and n is natural number, and corresponding according to algorithms selection
Physical network card sends function, message is sent, the corresponding transmission function of the physical network card that number is n is ethn_tx;
Specific calculation method is as follows:
When reaching a message, will pass, by load counter count=count+1;Message enters load balancing
Device module, load balancer obtain message number aggregate-value count and physical network card from load counter and load container respectively
Number m, passes through algorithm: Select=(count%m)+1;
If Selcet is equal to 1, calls eth1_tx to send function and send message;If Select=2, the hair of eth2_tx is called
Function is sent to send message;If Select is equal to n, calls ethn_tx to send function and send message;eth1_tx,eth2_tx,
Ethn_tx respectively corresponds the physical network card that uplink number is 1,2, n.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610913667.XA CN106375131B (en) | 2016-10-20 | 2016-10-20 | A kind of virtualization network uplink load-balancing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610913667.XA CN106375131B (en) | 2016-10-20 | 2016-10-20 | A kind of virtualization network uplink load-balancing method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106375131A CN106375131A (en) | 2017-02-01 |
CN106375131B true CN106375131B (en) | 2019-08-02 |
Family
ID=57894891
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610913667.XA Active CN106375131B (en) | 2016-10-20 | 2016-10-20 | A kind of virtualization network uplink load-balancing method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106375131B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110557340B (en) * | 2018-06-04 | 2023-04-07 | 中兴通讯股份有限公司 | Load balancing method, system and input device |
CN110611622B (en) * | 2018-06-15 | 2023-05-09 | 伊姆西Ip控股有限责任公司 | Method for load balancing, network interface card and computer readable medium |
CN109347656B (en) * | 2018-09-11 | 2022-02-18 | 郑州云海信息技术有限公司 | Method and system for configuring virtual host uplink network card |
CN113765816B (en) * | 2021-08-02 | 2023-12-15 | 阿里巴巴新加坡控股有限公司 | Flow control method, system, equipment and medium based on service grid |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101610283A (en) * | 2009-07-17 | 2009-12-23 | 浪潮电子信息产业股份有限公司 | A kind of method of dynamically balancing received load under network card binding state |
CN103051534A (en) * | 2012-11-20 | 2013-04-17 | 杭州迪普科技有限公司 | Message processing method and device |
CN103067976A (en) * | 2013-02-07 | 2013-04-24 | 深圳市磊科实业有限公司 | Load balancing method of multi-interface wireless access point and implementation device |
CN103188753A (en) * | 2011-12-27 | 2013-07-03 | 中国移动通信集团山东有限公司 | Method and device for data transmission based on different links between heterogeneous networks |
CN103384255A (en) * | 2011-12-31 | 2013-11-06 | 华为数字技术(成都)有限公司 | Load balancing method, server and system for virtual machine cluster |
CN104079486A (en) * | 2013-03-28 | 2014-10-01 | 国际商业机器公司 | Gateway and method of transmitting data through gateway |
CN104883264A (en) * | 2014-02-27 | 2015-09-02 | 中国科学院声学研究所 | Method and system of virtualizing network cards on network processing platform |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040024904A1 (en) * | 2002-07-31 | 2004-02-05 | Dimambro Francesco R. | Load balancing packet transmission among multiple transmit rings |
US20120230194A1 (en) * | 2011-03-11 | 2012-09-13 | Broadcom Corporation | Hash-Based Load Balancing in Large Multi-Hop Networks with Randomized Seed Selection |
-
2016
- 2016-10-20 CN CN201610913667.XA patent/CN106375131B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101610283A (en) * | 2009-07-17 | 2009-12-23 | 浪潮电子信息产业股份有限公司 | A kind of method of dynamically balancing received load under network card binding state |
CN103188753A (en) * | 2011-12-27 | 2013-07-03 | 中国移动通信集团山东有限公司 | Method and device for data transmission based on different links between heterogeneous networks |
CN103384255A (en) * | 2011-12-31 | 2013-11-06 | 华为数字技术(成都)有限公司 | Load balancing method, server and system for virtual machine cluster |
CN103051534A (en) * | 2012-11-20 | 2013-04-17 | 杭州迪普科技有限公司 | Message processing method and device |
CN103067976A (en) * | 2013-02-07 | 2013-04-24 | 深圳市磊科实业有限公司 | Load balancing method of multi-interface wireless access point and implementation device |
CN104079486A (en) * | 2013-03-28 | 2014-10-01 | 国际商业机器公司 | Gateway and method of transmitting data through gateway |
CN104883264A (en) * | 2014-02-27 | 2015-09-02 | 中国科学院声学研究所 | Method and system of virtualizing network cards on network processing platform |
Non-Patent Citations (1)
Title |
---|
"基于Linux的多网卡负载均衡技术";彭海云; 李亚;《实验室研究与探索》;20120915;第31卷(第9期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN106375131A (en) | 2017-02-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106375131B (en) | A kind of virtualization network uplink load-balancing method | |
CN104184813B (en) | The load-balancing method and relevant device and group system of virtual machine | |
CN104283939B (en) | For flexibly flowing the device of unloading, method and non-transitory computer-readable media | |
US9231869B2 (en) | Ensuring predictable and quantifiable networking performance | |
US11394649B2 (en) | Non-random flowlet-based routing | |
US10261840B2 (en) | Controlling virtual machine density and placement distribution in a converged infrastructure resource pool | |
US9258257B2 (en) | Direct memory access rate limiting in a communication device | |
US8429282B1 (en) | System and method for avoiding system overload by maintaining an ideal request rate | |
US8392558B1 (en) | System and method for determining overload state for service requests | |
CN107689925A (en) | Load balance optimization method and device based on cloud monitoring | |
US9762495B1 (en) | Weighted distribution across paths of degraded quality | |
US20180004693A1 (en) | Graphics processing unit (gpu) as a programmable packet transfer mechanism | |
US20180357086A1 (en) | Container virtual switching | |
CN105915462B (en) | A kind of symmetry RSS circuits towards TCP sessions | |
CN104348694A (en) | Network interface card with virtual switch and traffic flow policy enforcement | |
CN106027323A (en) | Technologies for gpu assisted network traffic monitoring and analysis | |
US10581680B2 (en) | Dynamic configuration of network features | |
US11301020B2 (en) | Data center power management | |
CN109218216A (en) | Chain polymerization traffic distribution method, device, equipment and storage medium | |
CN110022269A (en) | Processing method of communication data, device and equipment | |
Kaur et al. | Packet optimization of software defined network using lion optimization | |
US20220046647A1 (en) | Logical channel management in a communication system | |
US20230300073A1 (en) | Managing bandwith in fibre channel over internet protocol communication channels | |
da Silva et al. | A Dynamic Load Balancing Algorithm for Data Plane Traffic. | |
Adami et al. | Virtual machines migration in a cloud data center scenario: An experimental analysis |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |