CN1484413A - Method for realizing polylink flow equalization - Google Patents
Method for realizing polylink flow equalization Download PDFInfo
- Publication number
- CN1484413A CN1484413A CNA021370796A CN02137079A CN1484413A CN 1484413 A CN1484413 A CN 1484413A CN A021370796 A CNA021370796 A CN A021370796A CN 02137079 A CN02137079 A CN 02137079A CN 1484413 A CN1484413 A CN 1484413A
- Authority
- CN
- China
- Prior art keywords
- flow
- port
- address
- polylink
- packet
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
This invention discloses a method for realizing balance of aggregate link flow including the following steps: receiving a single broadcast data packet, if there are matched items in the address list, then they are forwarded if there is no matched address item, it is sent to an address study module which forecasts the future flow of each port according to its historical flow information to forward the said data packet via the port with the least flow in the future and renew the address list, broadly applied in an aggregate link bound by multiple physical links and a logic link between network devices, overcoming the shortcoming of unbanlanced loading for different ports decided according to certain bits of a data packet address.
Description
Technical field
The present invention relates between the network equipment method that many physical links are bundled into the polylink flow equalization of a logical links.
Background technology
Switch and switch can use the physical link more than two to connect between switch and the server.The benefit of doing like this is to have increased bandwidth, and the bandwidth of link aggregation equals the summation of the physical link bandwidth of polymerization.High reliability, when wherein certain bar physical link broke down, switch was discerned the inefficacy link automatically and is stopped to transmit information from this link, selected other links to proceed to transmit simultaneously.One or more physical link lost efficacy in the trunk, can not influence the connectedness of two equipment, and just link bandwidth descends along with the increase of stale link way, so the link aggregation technology can improve the reliability of trunk.Load balancing, link aggregation technology have a frame distribution algorithms, can carry out packet according to the flow of each physical link selectively and send, and make the load balancing of each link.Physical connection can dispose or reshuffle when changing fast.
Switch generally has a plurality of ports, and its major function is the packet that a port receives to be forwarded in the suitable port of another one go.The basic principle that realizes the exchange of packet is such: store data form and a study module of being called address table in the switch.Address table comprises an address entries and a port item at least, and the packet that is indicated as a certain destination address should send from corresponding port.Switch at first extracts destination address information after receiving a unicast packet from a port, and inquiry in the table of location if can find corresponding list item, is then transmitted this unicast packet according to corresponding port inwardly, handles otherwise press broadcast packet.Meanwhile study module can be according to the source address that receives bag and source port study and scheduler list item.
The scheme that realizes link aggregation has been described at United States Patent (USP) 6049528.An address table is arranged in equipment, after receiving a packet, at first search address table,, then transmit this packet according to the corresponding port in the address table if there is corresponding list item in the address table according to destination address.If do not have corresponding list item, then handle according to broadcast packet.Address list item then is that the rule learning according to general switch obtains.Application No. is to have described similar system in 20020012340 and 20020010791 explanations, and the rule that its exchange is transmitted is the partial information according to the packet that receives, and promptly which port a few positions of destination address are decided from and transmitted.
All there is a problem in above-described method, promptly all is that several bits according to the address decide this address and which port are interrelated during address learning.This method just might make that the certain port flow is bigger, and other port flow is very little, produces the phenomenon of flow skewness weighing apparatus between each port.
Summary of the invention
The technical problem of solution of the present invention has provided a kind of method that realizes polylink flow equalization, has overcome to decide according to the destination address that receives packet merely to occur the uneven phenomenon of flow in the link aggregation that exists in the way of address list item.
The method of realization polylink flow equalization of the present invention comprises following treatment step:
The following flow of prediction port specifically comprises following treatment step in the described step 3:
A) adopt the regularly method of inquiry, the port that reads accumulative total in the counter every T second sends byte number, supposes that the result of the j time inquiry of port i is Cij;
B) calculate the j time preceding T of inquiry of port i Mean Speed of second, computing formula is:
C) adopt the weighted average of several time period flows of past to estimate following flow.
If adopt the following T of the volume forecasting of two time periods Mean Speed of second, predictor formula is:
S '
I, j=α S
I, j+ (1-α) S
I, j-10≤α≤1 wherein
α represents weight, and maximum is 1, the big more firm T in the past of the dependence more flow of second of then predicting.
Can choose T=5 second generally speaking, α is 0.8.
Adopt the method for realization polylink flow equalization of the present invention, can overcome in the past decides the forwarding port may make the unbalanced shortcoming of different port load according to a few positions of data packet addressed, make when transmitting packet more balancedly dispense flow rate on physical link, the bandwidth of utilizing physical link to provide more fully.This method also has simply, fireballing advantage.
Description of drawings
Fig. 1 is the situation that two systems (such as between two switches or between the switch-to-server) connect by many physical links.
Fig. 2 is a functional module that port has of switch or server.
Fig. 3 is that port is handled the flow process that receives packet.
Fig. 4 is the address learning flow process of the present invention's method of realizing polylink flow equalization.
Embodiment
Below in conjunction with accompanying drawing basic principle of the present invention is described.
Fig. 1 is the situation that two systems (such as between two switches or between the switch-to-server) connect by many physical links.The port one of system A to port 4 links to each other with port 4 to the port one of the B of system, and these 4 link aggregations are a logical links.Non-polymeric port does not draw in this figure.
The functional module of having drawn a port among Fig. 2.The port receiver module is responsible for receiving packet from port.The address learning module is carried out address learning and scheduler table according to the source address and the receiving port number that receive packet.Address table has then write down the result of address learning, shows with a certain address to be which port the packet of destination address should forward from.Under the situation that TRUNKING (relaying) is not set, the CPU software processing module does not participate in the address learning process.Packet forward module is according to the destination address inquire address table that receives packet and make forwarding decision.The port sending module is realized the transmission of packet.
Fig. 3 is that aggregation port is handled the flow process that receives packet.A port is after being set to aggregation port, and the address learning module no longer works.Any unicast packet that receives is delivered to packet forward module earlier, and packet forward module is inquired about in address table according to the destination address of bag, if found the list item of coupling in address table, then carries out normal packet and transmits.If there is not the list item of coupling in the address table, then broadcasts processing, and address message is sent to the CPU software processing module carry out address learning.The software address study module comes the scheduler table with the result of study.
Fig. 4 is the address learning flow process of the present invention's method of realizing polylink flow equalization.In order to realize the purpose of flow equalization, at first predict each port flow in future, and determine the result of study according to the flow of prediction.CPU receives address learning message, and this address message comprises information such as the source address of the packet that receives and receiving port number.Inquire about the flow of all available ports in the aggregation group, and predict the flow of following port, selecting a port conduct of flow minimum is the forwarding port of the packet of destination address with this address, and address table is upgraded.
Introduce the implementation method of the following port flow of prediction below more in detail:
All have the ability of port flow statistics in the switch, the hardware of switch provides some cycle counters of 32, and software can know that by read counter port accumulative total has sent the data of how many bytes.Can adopt the regularly method of inquiry, inquire about counter in second, suppose that the result of the j time inquiry of port i is Cij every T.
The Mean Speed of T second is before the port i the j time inquiry:
The flow in common port future has bigger correlation with flow in the past, so we can adopt over the weighted averages of flow of several time periods and estimate the flow in future.In the formula below, we are with the Mean Speed of the following T of volume forecasting second of two time periods in the past:
S '
I, j=α S
I, j+ (1-α) S
I, j-10≤α≤1 wherein
α represents weight, and maximum is 1, the big more firm T in the past of the dependence more flow of second of then predicting.To after the following flow of front port is made a prediction, when occurring a new address at every turn and need learn, we come the scheduler list item according to address itself, but the port of this new address and flow minimum is connected, and the scheduler table at us.In above-mentioned formula, we choose T=5 second, and α 0.8 generally can obtain effect preferably.
More than describe the method that adopts method of the present invention many physical links between two equipment to aggregate into a logical links and realize load balancing in detail.
Claims (4)
1. method that realizes polylink flow equalization comprises following treatment step:
Step 1 receives a unicast packet, if the list item that is complementary is arranged in the address table then transmit;
Step 2 is if the address entries that is not complementary in the address table then is sent to the address learning module;
Step 3, the address learning module is predicted each port flow in future according to the port history flow information;
Step 4 is transmitted this packet by the port of following flow minimum, and the scheduler table.
2. the method for realization polylink flow equalization according to claim 1 is characterized in that, the following flow of prediction port specifically comprises following treatment step in the described step 3:
A) adopt the regularly method of inquiry, the port that reads accumulative total in the counter every T second sends byte number, supposes that the result of the j time inquiry of port i is Cij;
B) calculate the j time preceding T of inquiry of port i Mean Speed of second, computing formula is:
C) adopt the weighted average of several time period flows of past to estimate following flow.
3. the method for realization polylink flow equalization according to claim 2 is characterized in that, can adopt the Mean Speed of the following T of volume forecasting second of two time periods, and predictor formula is:
S '
I, j=α S
I, j+ (1-α) S
I, j-10≤α≤1 wherein
α represents weight, and maximum is 1, the big more firm T in the past of the dependence more flow of second of then predicting.
4. according to the method for claim 2 or 3 described realization polylink flow equalizations, it is characterized in that can choose T=5 second, α is 0.8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB021370796A CN100336363C (en) | 2002-09-18 | 2002-09-18 | Method for realizing polylink flow equalization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB021370796A CN100336363C (en) | 2002-09-18 | 2002-09-18 | Method for realizing polylink flow equalization |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1484413A true CN1484413A (en) | 2004-03-24 |
CN100336363C CN100336363C (en) | 2007-09-05 |
Family
ID=34146847
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB021370796A Expired - Fee Related CN100336363C (en) | 2002-09-18 | 2002-09-18 | Method for realizing polylink flow equalization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100336363C (en) |
Cited By (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100359879C (en) * | 2004-07-22 | 2008-01-02 | 阿尔卡特公司 | Local area network with one or more virtual groups of core equipments appropriate for layer 2 switching |
CN100417142C (en) * | 2005-12-22 | 2008-09-03 | 华为技术有限公司 | Method for average distributing interface flow at multi network processor engines |
CN100417133C (en) * | 2005-01-17 | 2008-09-03 | 华为技术有限公司 | Ethernet link converging method |
CN100426773C (en) * | 2005-11-30 | 2008-10-15 | 中兴通讯股份有限公司 | Method for equalizing port flow while multiple-MAC-port inter-connecting |
CN100438498C (en) * | 2005-01-17 | 2008-11-26 | 华为技术有限公司 | Ethernet device stacking method and its system |
CN100466620C (en) * | 2006-06-30 | 2009-03-04 | 南京联创科技股份有限公司 | Loading balance method based on data flow in large scale paralle processing of mass data |
CN100490419C (en) * | 2003-01-14 | 2009-05-20 | 株式会社日立制作所 | Information packet conmunication method and information packet communication device |
CN101170512B (en) * | 2007-11-21 | 2010-04-21 | 中兴通讯股份有限公司 | Message service processing method |
CN101286903B (en) * | 2008-05-06 | 2010-09-15 | 北京锐安科技有限公司 | Method for enhancing integrity of sessions in network audit field |
CN101009662B (en) * | 2007-01-31 | 2010-12-22 | 杭州华三通信技术有限公司 | Message processing method, system and device based on the load balance technology |
CN101056257B (en) * | 2006-04-14 | 2011-03-02 | 中兴通讯股份有限公司 | Method and system for implementing the link aggregation and protection switching |
CN101227402B (en) * | 2008-02-20 | 2011-03-23 | 杭州华三通信技术有限公司 | Method and apparatus for sharing polymerization link circuit flow |
CN101453368B (en) * | 2007-12-03 | 2012-02-29 | 华为技术有限公司 | Method, system and equipment for Internet IP address classification and bandwidth prediction |
US8228929B2 (en) | 2008-10-24 | 2012-07-24 | Juniper Networks, Inc. | Flow consistent dynamic load balancing |
CN102870382A (en) * | 2012-06-29 | 2013-01-09 | 华为技术有限公司 | Forwarding method, receiving method, first router an second router |
CN103023804A (en) * | 2012-12-18 | 2013-04-03 | 北京星网锐捷网络技术有限公司 | Method, device and network equipment for polymerization chain self-adaption flow load balance |
CN103155489A (en) * | 2010-10-15 | 2013-06-12 | 日本电气株式会社 | Switching system, and monitoring centralization management method |
CN105162698A (en) * | 2015-10-10 | 2015-12-16 | 浪潮(北京)电子信息产业有限公司 | Method and device for cloud server to adjust SDN network paths based on memory model |
CN105703976A (en) * | 2016-04-11 | 2016-06-22 | 浪潮集团有限公司 | Switch port state monitoring method |
CN106547481A (en) * | 2016-09-29 | 2017-03-29 | 浙江宇视科技有限公司 | A kind of data method for pre-distributing and equipment |
US9614768B2 (en) | 2011-04-06 | 2017-04-04 | Hewlett Packard Enterprise Development Lp | Method for traffic load balancing |
CN111010343A (en) * | 2019-12-18 | 2020-04-14 | 锐捷网络股份有限公司 | Method, device, network equipment and storage medium for forwarding multicast message |
CN113645145A (en) * | 2021-08-02 | 2021-11-12 | 迈普通信技术股份有限公司 | Load balancing method and device, network equipment and computer readable storage medium |
WO2022022558A1 (en) * | 2020-07-31 | 2022-02-03 | 中兴通讯股份有限公司 | Traffic allocation method and apparatus for aggregated link, and optical line terminal and storage medium |
CN114422416A (en) * | 2021-12-30 | 2022-04-29 | 广东中兴新支点技术有限公司 | Multilink aggregated data transmission method, system, device and storage medium |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6049528A (en) * | 1997-06-30 | 2000-04-11 | Sun Microsystems, Inc. | Trunking ethernet-compatible networks |
DE60034320T2 (en) * | 1999-03-17 | 2007-12-20 | Broadcom Corp., Irvine | METHOD FOR AVOIDING NON-EQUINENT FRAMEWORK IN A NETWORK TRANSMISSION AGENT |
-
2002
- 2002-09-18 CN CNB021370796A patent/CN100336363C/en not_active Expired - Fee Related
Cited By (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100490419C (en) * | 2003-01-14 | 2009-05-20 | 株式会社日立制作所 | Information packet conmunication method and information packet communication device |
US7706271B2 (en) | 2003-01-14 | 2010-04-27 | Hitachi, Ltd. | Method of transmitting packets and apparatus of transmitting packets |
CN100359879C (en) * | 2004-07-22 | 2008-01-02 | 阿尔卡特公司 | Local area network with one or more virtual groups of core equipments appropriate for layer 2 switching |
CN100438498C (en) * | 2005-01-17 | 2008-11-26 | 华为技术有限公司 | Ethernet device stacking method and its system |
CN100417133C (en) * | 2005-01-17 | 2008-09-03 | 华为技术有限公司 | Ethernet link converging method |
CN100426773C (en) * | 2005-11-30 | 2008-10-15 | 中兴通讯股份有限公司 | Method for equalizing port flow while multiple-MAC-port inter-connecting |
CN100417142C (en) * | 2005-12-22 | 2008-09-03 | 华为技术有限公司 | Method for average distributing interface flow at multi network processor engines |
CN101056257B (en) * | 2006-04-14 | 2011-03-02 | 中兴通讯股份有限公司 | Method and system for implementing the link aggregation and protection switching |
CN100466620C (en) * | 2006-06-30 | 2009-03-04 | 南京联创科技股份有限公司 | Loading balance method based on data flow in large scale paralle processing of mass data |
CN101009662B (en) * | 2007-01-31 | 2010-12-22 | 杭州华三通信技术有限公司 | Message processing method, system and device based on the load balance technology |
CN101170512B (en) * | 2007-11-21 | 2010-04-21 | 中兴通讯股份有限公司 | Message service processing method |
CN101453368B (en) * | 2007-12-03 | 2012-02-29 | 华为技术有限公司 | Method, system and equipment for Internet IP address classification and bandwidth prediction |
CN101227402B (en) * | 2008-02-20 | 2011-03-23 | 杭州华三通信技术有限公司 | Method and apparatus for sharing polymerization link circuit flow |
CN101286903B (en) * | 2008-05-06 | 2010-09-15 | 北京锐安科技有限公司 | Method for enhancing integrity of sessions in network audit field |
US8228929B2 (en) | 2008-10-24 | 2012-07-24 | Juniper Networks, Inc. | Flow consistent dynamic load balancing |
CN101729402B (en) * | 2008-10-24 | 2013-11-20 | 丛林网络公司 | Flow consistent dynamic load balancing |
US8446916B2 (en) | 2008-10-24 | 2013-05-21 | Juniper Networks, Inc. | Flow consistent dynamic load balancing |
CN103155489A (en) * | 2010-10-15 | 2013-06-12 | 日本电气株式会社 | Switching system, and monitoring centralization management method |
CN103155489B (en) * | 2010-10-15 | 2016-02-17 | 日本电气株式会社 | Switch system and supervision centralized control method |
US10623314B2 (en) | 2010-10-15 | 2020-04-14 | Nec Corporation | Switch system, and monitoring centralized control method |
US9614768B2 (en) | 2011-04-06 | 2017-04-04 | Hewlett Packard Enterprise Development Lp | Method for traffic load balancing |
CN102870382A (en) * | 2012-06-29 | 2013-01-09 | 华为技术有限公司 | Forwarding method, receiving method, first router an second router |
CN102870382B (en) * | 2012-06-29 | 2014-12-03 | 华为技术有限公司 | Forwarding method, receiving method, first router an second router |
CN103023804B (en) * | 2012-12-18 | 2015-08-19 | 北京星网锐捷网络技术有限公司 | A kind of aggregated links adaptive flow load-balancing method, device and the network equipment |
CN103023804A (en) * | 2012-12-18 | 2013-04-03 | 北京星网锐捷网络技术有限公司 | Method, device and network equipment for polymerization chain self-adaption flow load balance |
CN105162698A (en) * | 2015-10-10 | 2015-12-16 | 浪潮(北京)电子信息产业有限公司 | Method and device for cloud server to adjust SDN network paths based on memory model |
CN105162698B (en) * | 2015-10-10 | 2019-03-12 | 浪潮(北京)电子信息产业有限公司 | Method and apparatus of the Cloud Server based on memory models adjustment SDN network path |
CN105703976B (en) * | 2016-04-11 | 2018-11-09 | 浪潮集团有限公司 | A kind of monitoring method of switch port state |
CN105703976A (en) * | 2016-04-11 | 2016-06-22 | 浪潮集团有限公司 | Switch port state monitoring method |
CN106547481A (en) * | 2016-09-29 | 2017-03-29 | 浙江宇视科技有限公司 | A kind of data method for pre-distributing and equipment |
CN106547481B (en) * | 2016-09-29 | 2020-04-10 | 浙江宇视科技有限公司 | Data pre-distribution method and equipment |
CN111010343A (en) * | 2019-12-18 | 2020-04-14 | 锐捷网络股份有限公司 | Method, device, network equipment and storage medium for forwarding multicast message |
WO2022022558A1 (en) * | 2020-07-31 | 2022-02-03 | 中兴通讯股份有限公司 | Traffic allocation method and apparatus for aggregated link, and optical line terminal and storage medium |
CN113645145A (en) * | 2021-08-02 | 2021-11-12 | 迈普通信技术股份有限公司 | Load balancing method and device, network equipment and computer readable storage medium |
CN114422416A (en) * | 2021-12-30 | 2022-04-29 | 广东中兴新支点技术有限公司 | Multilink aggregated data transmission method, system, device and storage medium |
CN114422416B (en) * | 2021-12-30 | 2023-07-21 | 广东中兴新支点技术有限公司 | Method, system, device and storage medium for transmitting multilink aggregated data |
Also Published As
Publication number | Publication date |
---|---|
CN100336363C (en) | 2007-09-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100336363C (en) | Method for realizing polylink flow equalization | |
CN101668005B (en) | Data transmission accelerating engine method based on multiple access passages of transmitting end | |
CN1185829C (en) | Method of controlling Ethernet data flow quantity on synchronous numerical system transmission net | |
US20070053294A1 (en) | Network load balancing apparatus, systems, and methods | |
CN101170512B (en) | Message service processing method | |
US7680039B2 (en) | Network load balancing | |
CN1913486A (en) | Method and device for strengthening safety of protocol message | |
CN102035751A (en) | Data transmission method and equipment | |
US6009092A (en) | LAN switch architecture | |
CN101867511A (en) | Pause frame sending method, associated equipment and system | |
Tai et al. | Making large scale deployment of RCP practical for real networks | |
CN1910868A (en) | Method and device for controlling alignment buffer | |
CN101115008A (en) | Credit based flow control system, apparatus and method | |
CN102377640A (en) | Message processing apparatus, message processing method and preprocessor | |
CN101030940A (en) | Method for equilibrating network-flow load and the third-level exchanger | |
CN100589477C (en) | Method for realizing data sending based on binding link | |
CN1443412A (en) | Method for increasing data throughput in communication system | |
CN101202634B (en) | Single board improving data utilization ratio and system and method of data transmission | |
CN111541749B (en) | Data communication method and system of embedded equipment and related equipment | |
CN111193971B (en) | Machine learning-oriented distributed computing interconnection network system and communication method | |
CN101911612A (en) | Network offloading with reduced packet loss | |
CN103079231A (en) | Device and method for multicast-to-unicast based on wireless local area network in AP (Access Point) | |
CN101883047B (en) | Data concurrency transmission method of multi-network interface device | |
CN1620063A (en) | Method, system, and apparatus for a credit based flow control in a computer system | |
CN1890928A (en) | Network, router device, route updating suppression method used for the same, and program thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20070905 Termination date: 20170918 |