CN101193061B - Multi-Qos-based traffic control method - Google Patents

Multi-Qos-based traffic control method Download PDF

Info

Publication number
CN101193061B
CN101193061B CN2006101657228A CN200610165722A CN101193061B CN 101193061 B CN101193061 B CN 101193061B CN 2006101657228 A CN2006101657228 A CN 2006101657228A CN 200610165722 A CN200610165722 A CN 200610165722A CN 101193061 B CN101193061 B CN 101193061B
Authority
CN
China
Prior art keywords
qos
flow
priority
flow rate
real
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.)
Expired - Fee Related
Application number
CN2006101657228A
Other languages
Chinese (zh)
Other versions
CN101193061A (en
Inventor
穆国燕
黄钧
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ZTE Corp
Original Assignee
ZTE Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by ZTE Corp filed Critical ZTE Corp
Priority to CN2006101657228A priority Critical patent/CN101193061B/en
Publication of CN101193061A publication Critical patent/CN101193061A/en
Application granted granted Critical
Publication of CN101193061B publication Critical patent/CN101193061B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses a flow rate control method based on multi Qos which relates to the flow rate control method in the communication system. The invention is made to solve the problem in prior art that flow rate control with different Qos cannot be realized, which causes un-smoothly flow rate. The invention can detect current real-time flow rate through the network processor. If the real-time flow exceeds the network bandwidth limit, the micro engine is set to discard message from the low level to the high level in turn according to the Qos priority levels and a lost packet proportion is calculated. If the real-time flow is lower than the network bandwidth limit, the micro engine is set but no packet is discarded. If the real-time flow rate of the Qos remains on the critical state, the flow rate to be transmitted and the flow rate to be discarded are calculated according to the collected flow rate; the micro engine transmits Qos message from the high level to the low level in turn according to the Qos priority levels and the setting of the network processor; messages transmission of Qos with the real-time flow rate on the critical state is controlled according to the proportion of the transmission flow rate and the discarding flow rate. Thus flow rate control is effective.

Description

Flow control methods based on multi-Qos
Technical field
The present invention relates to flow control methods in the communication system, relate in particular on a kind of network processing unit based on multi-Qos (Quality of Service, service quality) flow control methods.
Background technology
Increasing owing to the access user in the network service, and the actual network bandwidth can not satisfy increasing user's demand, user's priority may have nothing in common with each other again, in this case, need to guarantee the user's of high preferential group the smoothness of network, can not make actual network take place congested again and cause meaningless packet loss.Present flow control algorithm is a lot, as token bucket algorithm or token bucket algorithm or Fair Queue algorithm, but certain limitation is arranged all, can not well address the above problem, be the difference that can not embody Qos priority, or be to cause flow unsmooth, or produce certain time delay.
Summary of the invention
In order to solve problems of the prior art, the object of the present invention is to provide a kind of flow control methods based on multi-Qos, this method adopts different control strategies to different Qos, has guaranteed the level and smooth transmission of flow.
For achieving the above object, the present invention is based on the flow control methods of multi-Qos, comprise the steps:
(1) network processing unit detects current real-time traffic, if real-time traffic surpasses network bandwidth limitations, then the priority order from low to high according to Qos is provided with micro engine dropping packets successively, and calculates the packet loss ratio; If real-time traffic is lower than network bandwidth limitations, no longer packet loss of micro engine is set then; Real-time traffic is in the Qos of critical condition, then flow that need send according to the flow rate calculation of gathering and the flow that abandons;
(2) micro engine sends the message of Qos according to the setting of network processing unit according to the priority order from high to low of Qos; Real-time traffic is in the Qos of critical condition, then according to transmitted traffic and the message that abandons flow proportional control transmission Qos.
In the described step (1), micro engine dropping packets successively is set, and the method for calculating the packet loss ratio comprises according to the priority order from low to high of Qos:
(11) calculate redundant flow, the priority level of Qos is made as minimum, described redundant flow is the poor of current real-time traffic and network bandwidth limitations;
(12) if described redundant flow equals 0, the packet loss ratio that the Qos of this priority then is set is packet loss not;
If described redundant flow is greater than 0, the then relatively current transmitted traffic of other Qos of this level and the size of described redundant flow, if the current transmitted traffic of the Qos of this priority is less than or equal to described redundant flow, the packet loss ratio of Qos that this priority then is set is for all abandoning, and the value of the difference of the current transmitted traffic of the Qos of described redundant flow and this priority is composed to described redundant flow; If the current transmitted traffic of the Qos of this priority is greater than described redundant flow, the packet loss ratio that the Qos of this priority then is set is: the current of the current transmitted traffic of the Qos of this priority and the Qos of this priority abandons value/described redundant flow flow and that deduct the difference of described redundant traffic income, and the value of described redundant flow is put 0;
(13) if the priority level of Qos is a limit priority, then finish, otherwise the rank of Qos adds 1, turns to step (12).
Further, in the described step (11), the priority level of Qos is minimum to be 0.
In the described step (2), real-time traffic is in the Qos of critical condition, then according to transmitted traffic with abandon the method for message that flow proportional control sends Qos and adopt microcode proportional control algorithm, this algorithm comprises:
(21) whether the restriction proportion of judging the flow input of Qos changes, if then calculate the n value that sends the minimum scale section according to transmitted traffic s with the ratio that abandons flow d, thereby obtain in the minimum scale section transmitted traffic byte number s/n and abandon byte number d/n; Enter (22); If not, also enter (22); The wheel number that described n value sent for each second;
(22) whether the message length that judge to send less than the s/n of epicycle, if less than, then send message and add up transmitted traffic; If be not less than, enter (23);
(23) judge the message length abandon whether less than the d/n of epicycle, if less than, dropping packets and statistics abandon flow; If be not less than, enter (24);
(24) judge whether to flow-compensated compensation wheel, if then transmission and the byte number and the offset that do not abandon not add up, and enter (25); If not, then will be added to next round and send, and send message and add up transmitted traffic less than the byte number that sends and do not abandon;
(25) enter the compensation wheel and send, after the compensation wheel is finished dealing with, the remainder that sends or abandon is added to next round sends.
Further, the method that calculate to send the n value of minimum scale section according to flow proportional in the described step (21) is:
n=Min(s,d)/(1518*8)
Wherein (s, value d) is s to Min, d smaller's in the two value.
The inventive method to the transmitted traffic of different Qos with abandon flow proportional and control, can be good at reaching the effect of flow control by the network microprocessor, makes flow level and smooth, and can well embody the difference of different Q os.Realized ensureing under the high Qos, made full use of the purpose of the network bandwidth.
Description of drawings
Fig. 1 is the flow chart of the CPU part of specific embodiment of the invention coupling application program;
Fig. 2 is the specific embodiment of the invention microcode algorithm flow chart of transmitted traffic in proportion.
Embodiment
Below in conjunction with accompanying drawing the enforcement of technical scheme of the present invention is done and to be described in further detail, the present invention adopts the method that partly combines at network processing unit CPU part and micro engine to carry out flow control, shown in Figure 1 is the flow chart that upper layer network processor CPU semiconvection amount is controlled, and mainly comprises:
Upper layer network processor CPU regularly detects current real-time traffic, and this real-time traffic comprises flow that really sends out and the flow summation that abandons.When real-time traffic surpasses bandwidth constraints,, micro engine dropping packets successively is set according to Qos order from low to high; When real-time traffic is lower than bandwidth constraints, no longer packet loss of micro engine is set then; Real-time traffic is in Qos under the critical condition, the flow that need send and abandon according to the flow rate calculation of gathering, thus make flow controlled.
It specifically comprises:
The first step reads network bandwidth limitations (Restrict).
Second step, detect the timer expiry of flow, obtain transmitted traffic and abandon the traffic statistics data from the statistics district of micro engine; Calculate current transmitted traffic (SendFlux) and abandon flow (LostFlux), then real-time traffic is AllFlux=SendFlux+LostFlux, if AllFlux>Restrict then calculates the packet loss ratio; If AllFlux≤Restrict makes its no longer packet loss with the recovery of all Qos.
The step of wherein calculating the packet loss ratio comprises:
(2.1) difference of supposing current real-time traffic and bandwidth constraints is redundant flow OutofFlux, and the priority of the Qos of current detection is DetectQos, and lowest priority Qos is 0; Begin to detect from DetectQos=0, the current transmitted traffic of DetectQos is SendFluxQos, and abandoning flow is LostFluxQos;
(2.2) if OutofFlux=0 recovers DetectQos and makes its no longer packet loss;
If OutofFlux>0, compare the current transmitted traffic SendFluxQos of DetectQos and the size of redundant flow OutofFlux, if current transmitted traffic SendFluxQos is less than or equal to redundant flow OutofFlux, be SendFluxQos≤OutofFlux, then the transmission ratio of this Qos message is for all abandoning, the corresponding minimizing of redundant flow is promptly: OutofFlux=OutofFlux-SendFluxQos, if the transmitted traffic of current Qos, then is provided with the packet loss ratio greater than redundant flow:
(SendFluxQos+LostFluxQos-OutofFlux): (OutofFlux), and outofFlux put 0, i.e. outofFlux=0.
(2.3) if the priority level of Qos is a limit priority, then finish, otherwise DetectQos=DetectQos+1 forwards step (2.2) to, up to all Qos of traversal.
The processing of micro engine part comprises:
The requirement that transmit queue scheduler module in the micro engine is provided with according to upper layer network processor CPU when sending message at different Qos, at first sends the Qos of high priority, sends the Qos of low priority again.When transmission or dropping packets, the message byte number that sends or abandon according to the Qos statistics sends; To the Qos that sends by the Critical Control state, according to transmitted traffic and the message that abandons this Qos of flow proportional control transmission.Wherein, to the Qos that sends by the Critical Control state, adopt microcode proportional control algorithm according to transmitted traffic and the method for message that abandons this Qos of flow proportional control transmission.
Shown in Figure 2 is microcode proportional control algorithm flow:
This algorithm is according to the result of last data sampling, calculates the beat that sends and abandon, and needs the token number that sends and abandon in each beat, sends and dropping packets according to token number then.This token number is dynamically adjusted according to current real-time traffic, and compensation piecemeal, thereby the flow of output is almost equated with the flow that CPU is provided with.Mainly comprise the steps:
(A) whether the restriction proportion of judging the flow input of Qos changes, if then calculate the n value that sends the minimum scale section according to transmitted traffic s with the ratio that abandons flow d, thereby obtain in the minimum scale section transmitted traffic byte number s/n and abandon byte number d/n; Enter (B); If not, also enter (B);
Wherein, the n value is the wheel number that send each second, and (s, d)/(1518*8), (s, value d) is s to Min to n=Min, d smaller's in the two value.
(B) whether the message length that judge to send less than the s/n of epicycle, if less than, then send message and add up transmitted traffic; If be not less than, enter (C);
(C) judge the message length abandon whether less than the d/n of epicycle, if less than, dropping packets and statistics abandon flow; If be not less than, enter (D);
(D) judge whether to flow-compensated compensation wheel, if then transmission and the byte number and the offset that do not abandon not add up, and enter (E); If not, then will be added to next round and send, and send message and add up transmitted traffic less than the byte number that sends and do not abandon;
(E) enter the compensation wheel and send, after the compensation wheel is finished dealing with, the remainder that sends or abandon is added to next round sends.
The specific implementation of above-mentioned steps is:
To flow control the time, the flow that expectation sends is thin more good more with the expectation flow granularity of input, but that this granularity is subject to the bag of Ethernet is long, and according to RFC894, the long scope of bag is 64-1518, and it is 1518byte that the granularity minimum is controlled in selection.Because input is to answer transmitted traffic (s) and abandon flow (d), the bag that just sends and abandon corresponding under a second is long, the transmission in a second can be abandoned like this and be divided into the n wheel, every Bao Changwei s/n that takes turns transmission, the Bao Changwei d/n that abandons, n=k/ (1518*8) wherein, k is s, the minimum value of d.Because in the process of transmitting of reality, the Bao Changyu s/n that sends and abandon, d/n often have remainder, every take turns send finish after, this remainder is added up into next round, by way of compensation.After sending n wheel, the flow of transmission is n*[s/n], n*[d/n], compensate the difference between itself and s and the d.
(a) input flow rate that obtains, its ratio are s: d, and s wherein, d represent the transmitted traffic of this moment, and d represents the flow that abandons of this moment;
(b) compare s and d, getting k is s, the minimum value among the d;
(c) n that obtains transmission in 1 second takes turns n=k/ (1518*8);
(d) obtain the every byte number s/n that sends and abandon, d/n of taking turns;
(e) whether the byte number that judge to send arrives the s/n that this is taken turns, if not then send, if surpass then enter the judgement that abandons that this takes turns;
(f) judge whether the message abandon arrives the d/n that this is taken turns,, finish if surpass then should wheel send if not then abandon;
(g) one take turns send to finish after, with s/n, d/n adds up with actual byte number surplus that send and that abandon and sends into next round;
(h) n wheel send receive after, with s-n*[s/n] and d-n*[d/n] remainder add and send and abandon link, the error in 1 second is compensated.
The error of algorithm: the deviant of error is s-n*[s/n to the maximum in a second]+Δ (Δ is last round of offset), obvious Δ<1518*8, s-n*[s/n]<n; N=k/ (1518*8)<100*1024*1024/ (1518*8)<1518*8, it is long to that is to say that error is no more than the bag of two 1518 bytes when maximum, and because the continuous compensation of error does not have diversity.
This method adopts different control strategies to different Qos on the basis of sampling, and guarantees the level and smooth transmission of flow as far as possible; Realize simply, effectively being particularly useful for carrying out in the network processing unit flow control of multi-Qos.

Claims (2)

1. the flow control methods based on multi-Qos comprises the steps:
(1) network processing unit detects current real-time traffic, if real-time traffic surpasses network bandwidth limitations, then the priority order from low to high according to Qos is provided with micro engine dropping packets successively, and calculates the packet loss ratio; If real-time traffic is lower than network bandwidth limitations, no longer packet loss of micro engine is set then; Real-time traffic is in the Qos of critical condition, then flow that need send according to the flow rate calculation of gathering and the flow that abandons;
Described priority order from low to high according to Qos is provided with micro engine dropping packets successively, and the method for calculating packet loss ratio comprises:
(11) calculate redundant flow, the priority level of Qos is made as minimum, described redundant flow is the poor of current real-time traffic and network bandwidth limitations;
(12) if described redundant flow equals 0, the packet loss ratio that the Qos of this priority then is set is packet loss not;
If described redundant flow is greater than 0, the then relatively current transmitted traffic of other Qos of this level and the size of described redundant flow, if the current transmitted traffic of the Qos of this priority is less than or equal to described redundant flow, the packet loss ratio of Qos that this priority then is set is for all abandoning, and the value of the difference of the current transmitted traffic of the Qos of described redundant flow and this priority is composed to described redundant flow; If the current transmitted traffic of the Qos of this priority is greater than described redundant flow, the packet loss ratio that the Qos of this priority then is set is: the current of the current transmitted traffic of the Qos of this priority and the Qos of this priority abandons value/described redundant flow flow and that deduct the difference of described redundant traffic income, and the value of described redundant flow is put 0;
(13) if the priority level of Qos is a limit priority, then finish, otherwise the rank of Qos adds 1, turns to step (12);
(2) micro engine sends the message of Qos according to the setting of network processing unit according to the priority order from high to low of Qos; Real-time traffic is in the Qos of critical condition, then according to transmitted traffic with abandon the message that flow proportional control sends Qos;
Described according to transmitted traffic and the method for message employing microcode proportional control algorithm that abandons flow proportional control transmission Qos, this algorithm comprises:
(21) whether the restriction proportion of judging the flow input of Qos changes, if then calculate the n value that sends the minimum scale section according to transmitted traffic s with the ratio that abandons flow d, thereby obtain in the minimum scale section transmitted traffic byte number s/n and abandon byte number d/n; Enter (22); If not, also enter (22); Described n value is the wheel number that send each second, n=Min (s, d)/(1518*8), wherein Min (s, value d) is s, d smaller's in the two value; Restriction proportion is the Qos that real-time traffic is in critical condition, transmitted traffic and the ratio that abandons flow;
(22) whether the message length that judge to send less than the s/n of epicycle, if less than, then send message and add up transmitted traffic; If be not less than, enter (23);
(23) judge the message length abandon whether less than the d/n of epicycle, if less than, dropping packets and statistics abandon flow; If be not less than, enter (24);
(24) judge whether to flow-compensated compensation wheel, if then transmission and the byte number and the offset that do not abandon not add up, and enter (25); If not, then will be added to next round and send, and send message and add up transmitted traffic less than the byte number that sends and do not abandon;
(25) enter the compensation wheel and send, after the compensation wheel is finished dealing with, the remainder that sends or abandon is added to next round sends.
2. the flow control methods based on multi-Qos according to claim 1 is characterized in that, in the described step (11), the priority level of Qos is minimum to be 0.
CN2006101657228A 2006-12-14 2006-12-14 Multi-Qos-based traffic control method Expired - Fee Related CN101193061B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2006101657228A CN101193061B (en) 2006-12-14 2006-12-14 Multi-Qos-based traffic control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2006101657228A CN101193061B (en) 2006-12-14 2006-12-14 Multi-Qos-based traffic control method

Publications (2)

Publication Number Publication Date
CN101193061A CN101193061A (en) 2008-06-04
CN101193061B true CN101193061B (en) 2011-07-13

Family

ID=39487822

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2006101657228A Expired - Fee Related CN101193061B (en) 2006-12-14 2006-12-14 Multi-Qos-based traffic control method

Country Status (1)

Country Link
CN (1) CN101193061B (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101442486B (en) * 2008-12-24 2011-06-08 华为技术有限公司 Method and apparatus for distributing micro-engine
CN101945437B (en) * 2009-07-03 2013-08-28 华为技术有限公司 Method for determining traffic distribution strategy, access network equipment and route equipment
CN101616097B (en) * 2009-07-31 2012-05-23 中兴通讯股份有限公司 Method and system for managing output port queue of network processor
CN101800698B (en) * 2010-01-29 2012-02-01 华南理工大学 Flow limit system and method thereof based on network processor
WO2011147179A1 (en) * 2010-05-28 2011-12-01 烽火通信科技股份有限公司 Method for managing network service flow of ptn
CN101867580B (en) * 2010-06-03 2012-11-28 北京蓝汛通信技术有限责任公司 Method for allocating network flow and device
US8964538B2 (en) * 2010-09-24 2015-02-24 Broadcom Corporation Method and apparatus for supporting differentiated performance for multiple categories of packets in a passive optical network
US9148386B2 (en) * 2013-04-30 2015-09-29 Cisco Technology, Inc. Managing bandwidth allocation among flows through assignment of drop priority
CN104702531B (en) * 2013-12-10 2018-04-20 华为技术有限公司 The method and the network equipment that a kind of network apparatus jamming avoids
CN104052681B (en) * 2014-06-25 2018-09-21 中国联合网络通信集团有限公司 Flow control methods and device
CN105656807B (en) * 2015-12-25 2019-02-22 盛科网络(苏州)有限公司 A kind of network chip multi-channel data transmission method and transmitting device
CN115473819A (en) * 2022-08-30 2022-12-13 电信科学技术第十研究所有限公司 System and method for processing mass internet traffic based on dynamic rule driving

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2004036845A1 (en) * 2002-10-18 2004-04-29 Telefonaktiebolaget Lm Ericsson (Publ) Arrangements and method for controlling transmission of data bits
CN1607767A (en) * 2003-10-16 2005-04-20 中兴通讯股份有限公司 Flow-control method based on network processor
CN1658575A (en) * 2005-03-21 2005-08-24 北京北方烽火科技有限公司 Method for improving service quality in SGSN network processor
WO2005112347A2 (en) * 2004-05-05 2005-11-24 Cisco Technology, Inc. HIERARCHICAL QoS BEHAVIORAL MODEL
CN1859303A (en) * 2006-01-25 2006-11-08 华为技术有限公司 Dynamic flow control method based on end port

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2004036845A1 (en) * 2002-10-18 2004-04-29 Telefonaktiebolaget Lm Ericsson (Publ) Arrangements and method for controlling transmission of data bits
CN1607767A (en) * 2003-10-16 2005-04-20 中兴通讯股份有限公司 Flow-control method based on network processor
WO2005112347A2 (en) * 2004-05-05 2005-11-24 Cisco Technology, Inc. HIERARCHICAL QoS BEHAVIORAL MODEL
CN1658575A (en) * 2005-03-21 2005-08-24 北京北方烽火科技有限公司 Method for improving service quality in SGSN network processor
CN1859303A (en) * 2006-01-25 2006-11-08 华为技术有限公司 Dynamic flow control method based on end port

Also Published As

Publication number Publication date
CN101193061A (en) 2008-06-04

Similar Documents

Publication Publication Date Title
CN101193061B (en) Multi-Qos-based traffic control method
Sisalem et al. MLDA: A TCP-friendly congestion control framework for heterogeneous multicast environments
EP1780961B1 (en) Flow control for real-time data streams
CN109120544B (en) Transmission control method based on host end flow scheduling in data center network
JP4632874B2 (en) Communication terminal
CN102461093B (en) Method of managing a traffic load
CN103457871B (en) Based on the increasing window method of the congestion avoidance phase of deferred constraint in DCN
US20030214951A1 (en) Method for differentiating services and users in communication networks
Sisalem et al. LDA+: A TCP-friendly adaptation scheme for multimedia communication
JP2004538719A (en) Method for providing a non-linear, highly scalable increase-decrease congestion control mechanism
CN102291389A (en) Cross-layer congestion control method in satellite network
CN109873773B (en) Congestion control method for data center
CN112737964B (en) Transmission control method and system integrating push-pull semantics
CN112995048B (en) Blocking control and scheduling fusion method of data center network and terminal equipment
EP1526686B1 (en) Switching unit with scalable and QoS aware flow control
CN112822268B (en) Multi-service coexistence and IP layer packet scheduling method in industrial Internet of things
CN100568847C (en) Frame scheduling in differentiated services networks and buffer management unified algorithm
KR20090085906A (en) Routing processing system and the control method by priority order of logic interface
CN115665060A (en) Multi-path transmission scheduling method and device for heterogeneous network
CN102404208B (en) Stable network congestion control method with high-efficiency
TWI423623B (en) Network buffer management methods and computer program products
CN113347114A (en) Real-time streaming media transmission control method and device facing deadline sensing
Albuquerque et al. An end-to-end source-adaptive multi-layered multicast (SAMM) algorithm
Adhari et al. Eclipse: A new dynamic delay-based congestion control algorithm for background traffic
CN107302504B (en) Multipath transmission scheduling method and system based on virtual sending queue

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

Granted publication date: 20110713

Termination date: 20151214

EXPY Termination of patent right or utility model