CN1866927A - Information switching realizing system and method and scheduling algorithm - Google Patents
Information switching realizing system and method and scheduling algorithm Download PDFInfo
- Publication number
- CN1866927A CN1866927A CNA2006100765066A CN200610076506A CN1866927A CN 1866927 A CN1866927 A CN 1866927A CN A2006100765066 A CNA2006100765066 A CN A2006100765066A CN 200610076506 A CN200610076506 A CN 200610076506A CN 1866927 A CN1866927 A CN 1866927A
- Authority
- CN
- China
- Prior art keywords
- packet
- input port
- formation
- algorithm
- crosspoint
- 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
Abstract
The invention relates to a system for realizing information communication and relative distribution method. Wherein, said system comprises an input port and a communicate unit; the input port is used to input data pack; said communicate unit is used to communicate date between the input port and the output port; the input port has delay queue; the input port divides each delay queue and selects data pack to be transferred to the communicate unit; the communicate unit stores the data pack and selects data pack to be transferred to the output port. The invention can divide the service data, and increase the insulation level according to the service demand.
Description
Technical field
The present invention relates to the network communication field of information technology, relate in particular to a kind of system and method and dispatching algorithm that realizes information exchange.
Background technology
Switching fabric is the main architecture that current realization Fast Packet is transmitted, design philosophy in the past is to consider its versatility, the port that is switching fabric be towards the business that might arrive design, so switching fabric is also identical for the disposal ability that they provide.Mostly prior art is based on this performance of switching fabric to be studied, designed different cache policies, as (the OQ of output work queue, Output-Queued) the structure, (IQ of input queue, Input-Queued) structure, associating input/output queue (CIOQ, Combined Input-Output-Queued) structure etc.
With reference to Fig. 1, the switching system of prior art one is that all buffer queues all are arranged on the output port place in output work queue (OQ) structure, and this system requirements interconnected network can be delivered to destination interface with a plurality of bags simultaneously.When the bag of a plurality of input ports is gone to same destination interface simultaneously, be unfavorable for expansion, be not suitable for being directly used in high velocity environment.
With reference to Fig. 2, the switching system of prior art two is that all buffer queues all are arranged on the input port place in input queue (IQ) structure, and the bag of arrival at first is stored in the buffer queue, is dispatched by input port.When the bag of a plurality of formation heads will be gone to same output port, described technology can produce team's head (HOL, Head of Line) blocked, and can not make full use of link circuit resource.
With reference to Fig. 3, the switching system of prior art three is that associating input/output queue (CIOQ) structure cushions queuing respectively at input/output port.Though Shang-Tse Chuang etc. proves that minimum speed-up ratio only is required to be 2-1/N, because of too high based on the dispatching algorithm complexity of this kind structure proposition at present, so only have theory significance.
Yet all there is following defective in the switching system of above-mentioned prior art aspect caching mechanism:
1, the specific aim of port design is not strong, and isolation is poor;
2, each port all is towards all business, all business datums is handled, and business datum is not distinguished, and is unfavorable for Differentiated Services.
Present many dispatching algorithms are that the caching mechanism according to above-mentioned switching system proposes, and these caching mechanisms match with dispatching algorithm, have obtained good effect in actual applications, and main dispatching algorithm has following several:
But the algorithm based on priority is a kind of queue scheduling algorithm of simple relatively supporting area graduation service.In typical priority queueing (PQ, Priority Queuing), bag is sent into buffer memory in the different priority queries after through classification, and scheduler preferentially sends bag in the higher priority queues in strict accordance with priority order from high to low.In each priority query, bag is accepted scheduling according to the criterion of serving (FCFS) first earlier, and this scheduling is called strict priority queuing (SPQ, Strict PriorityQueuing) mechanism.The dispatching algorithm computation complexity of priority is low, and can be highest-priority queue provides low time delay to guarantee, but fairness is poor.
Broad sense processors sharing (GPS, Generalized Processor Sharing) algorithm is a kind of Utopian algorithm, and these algorithm hypothesis incoming traffic data can infinitely be segmented (Fluid-flow Mode).GPS can overcome fair-queueing (FQ, Fair Queuing) can not really realize fair service in elongated packet network defective, but this algorithm only has the value of theory analysis.
Packet-by-packet broad sense processors sharing (PGPS, Packet-by-Packet GPS) algorithm is the attainable approximate version of GPS, and it carries out in the mode of scheduling packet-by-packet, but this algorithm complex is big, is difficult to realize.
Weighted Fair Queuing (WFQ, Weighted Fair Queuing) algorithm is to propose by on the basis of bit poll (BBRR, Bit-by-bit Round Robin Service) algorithm, and BBRR only is a theoretical analysis method.For with BBRR algorithm practicability, people such as Demers have proposed the WFQ algorithm again, and when being polled to a certain formation, the bit number of dispatching from this formation is by the weights decision of this formation, but the weighted-fair-queuing complexity is big, is difficult to realize.
WRR (WRR, Weighted Round Robin) algorithm is cached to bag in the corresponding priority level formation, and each formation has different normalization weights, when being polled to this formation, sends a certain amount of data according to its weights.But the WRR algorithm can't accurately distribute the bandwidth ratio for elongated packet scheduling, can not accurately guarantee fairness.
Difference WRR (DWRR, Deficit Weighted Round Robin) algorithm can't distribute for elongated bag provides designated bandwidth in order to improve WRR.DWRR is putting in order under the prerequisite of packet scheduling, can distribute for different service types realizes predetermined long-term relative bandwidth, and reduce the computation complexity of WFQ.But DWRR algorithm priority characteristic is good inadequately, and aspect the low time delay business of support, its performance is than priority queueing difference.
According to the shortcoming of above-mentioned several dispatching algorithms as can be known, the defective of prior art is to satisfy priority simultaneously, algorithm complex, time delay, the requirement of four aspects of fairness.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of system and method and dispatching algorithm that realizes information exchange, the present invention can serve differentiation to business datum, simultaneously can strengthen isolation according to demand for services, and can satisfy priority, algorithm complex, time delay, the requirement of four aspects of fairness.
For solving the problems of the technologies described above, the invention provides a kind of system that realizes information exchange, this system is achieved in that
A kind of system that realizes information exchange, this system comprises input port and crosspoint, and described input port is used to import packet, and described crosspoint is used to realize the exchanges data of input port and output port;
At input port buffer queue is set, input port is divided each buffer queue, selects packet to be forwarded to crosspoint again, and crosspoint is preserved packet, selects packet to send to output port again.
Wherein, described buffer queue adopts the VOQ structure.
Wherein, described input port comprises scheduler, is used to judge dispatch service, the packet that described scheduler sends to crosspoint according to the difference WRR SPDWRR algorithm selection of supporting to seize.
Wherein, described crosspoint comprises output buffers formation and output scheduling device, and described output buffers formation is used to store the packet that outputs to same destination interface; The described output scheduling device packet that selection sends to output port according to the SPDWRR algorithm.
For solving the problems of the technologies described above, the invention provides a kind of method that realizes information exchange, this method may further comprise the steps:
A, input port inside are provided with buffer queue, and input port is divided each buffer queue;
B, input port select packet to send to crosspoint;
C, crosspoint storage packet select packet to send to output port again.
Wherein, input port is divided specifically each buffer queue and is comprised in the steps A:
Input port is provided with a formation for each output port, and the packet of input is cached in the corresponding formation according to its destination interface, further classifies the corresponding buffer queue of each service type again according to the demand for services of input packet.
Wherein, step B specifically comprises:
After input port was divided buffer queue, scheduler adopted dispatching algorithm to judge dispatch service according to demand for services, and selected packet to send to crosspoint.
Wherein, the algorithm of described scheduler employing is the SPDWRR algorithm.
Wherein, step C specifically comprises:
C1, crosspoint inside are provided with buffer queue, and crosspoint stores packet into the corresponding cache formation according to the destination interface of input packet;
C2, output scheduling device adopt dispatching algorithm to judge dispatch service according to demand for services, and select packet to send to output port.
Wherein, the algorithm of the employing of output scheduling device described in the step C2 is the SPDWRR algorithm.
For solving the problems of the technologies described above, the invention provides a kind of dispatching algorithm, this algorithm may further comprise the steps:
A, will import data qualification, the corresponding formation of every class has highest-priority queue Q0 in the formation;
Whether b, scheduler judge the difference of the balance counter C0 of high-priority queue Q0 and the packet length L0 of team more than or equal to 0, if Q0 obtains service, repeating step b is up to C0-L0<0, execution in step c again, otherwise direct execution in step c;
Whether the difference of c, the balance counter Ci of Qi that judges snoop queue and the packet length Li of team, changes and remove execution in step b, otherwise the operation of execution in step c is until Ci-Li<0 if Qi obtains service more than or equal to 0.
Wherein, the chain type storage organization is adopted in described formation.
Above technical scheme as can be known, the invention provides a kind of system and method for realizing information exchange, described system and method all is to adopt the two-stage type caching mechanism to realize information exchange, technique scheme is provided with buffer memory at input port, each formation to input port is segmented, be divided into different service queues by demand for services, can scheduling strategy be set at certain professional demand for services; Further, this technical scheme is provided with buffer queue at each output port in switching fabric inside, and this set of queues has not only been preserved the packet that outputs to same destination interface, and strengthens isolation by demand for services again.Guarantee to postpone (GD for focusing on simultaneously, Guaranteed-Delay), guarantee bandwidth (GB, Guaranteed-Bandwidth), (BE, Best-Effort) the class business provides solution, and is with good expansibility and lower computation complexity in best effort.
In addition, as shown from the above technical solution, the present invention also provides a kind of dispatching algorithm of special use, and the bag of high-priority queue preferentially obtains scheduling than the bag of Low Priority Queuing in the share scope, thereby satisfies the demand of priority; Further, high Low Priority Queuing obtains scheduling in turn in this algorithm, after the share of high-priority queue uses up,, cause " dying of hunger " because of can not get serving to avoid the bag in the Low Priority Queuing even also have the bag also should the schedules low priority formation in the formation.In addition, because this dispatching algorithm supports highest-priority queue to seize, so highest-priority queue can obtain than time delay littler under the DWRR algorithm, this dispatching algorithm has lower delay character simultaneously, and formation can obtain minimum time delay all the time under different loads.
Description of drawings
The switching system of Fig. 1 prior art one;
The switching system of Fig. 2 prior art two;
The switching system of Fig. 3 prior art three;
Fig. 4 system of the present invention detailed structure view;
Fig. 5 main structure of system figure of the present invention;
The caching mechanism schematic diagram of Fig. 6 system of the present invention;
Fig. 7 method flow diagram of the present invention;
The high-priority queue scheduling flow figure of Fig. 8 dispatching algorithm of the present invention;
The Low Priority Queuing scheduling flow figure of Fig. 9 dispatching algorithm of the present invention;
Figure 10 queue chain array schematic diagram.
Embodiment
The invention provides a kind of system that realizes information exchange, the basic thought of this system is: at input port buffer memory is set, adopt virtual output work queue structure, each VOQ is divided into different service queues by demand for services, each VOQ can be divided into a plurality of service types, and the scheduler of input port selects packet to transmit again; Packet carries out the secondary buffer memory according to the destination interface that wraps after leaving input port, is forwarded to output port by the output scheduling device in the switching fabric again.
According to above-mentioned basic thought, in conjunction with the accompanying drawings system of the present invention is described again.With reference to Fig. 5, Fig. 5 is main structure of system figure of the present invention, and described system comprises input port 200, crosspoint 300 and output port 400, wherein input port 200 comprises input-buffer formation 201 and scheduler 202, and crosspoint 300 comprises output buffers formation 301 and output scheduling device 302.
Below in conjunction with Fig. 6 the caching mechanism of system of the present invention is described in detail further, the first order, at input port buffer memory is set, adopt the VOQ structure, each VOQ formation to input port is segmented, be divided into different service queues by demand for services, be called fq, when a VOQ formation can be divided into p service type (p>1), then a VOQ formation comprises p fq formation, because each input port receives only certain business, so certain scheduling strategy can be set at the demand for services of this business.The second level at each output port, is provided with one group of buffer queue in asymmetric switching fabric inside, and this set of queues had both been preserved the packet that outputs to same destination interface, had strengthened isolation by demand for services again.Design the two-stage type caching mechanism thus, this mechanism comprises N input port and N output port, is provided with buffer queue at input port and switching fabric inside.In Fig. 6, input port is that i and output port are i.At input port i, divide the VOQ formation by the destination interface of input packet, for each VOQ formation, further classify according to the demand for services of bag again, as formation VOQi, j is divided into q fq formation by demand for services, promptly from fqi, j, 1 to fqi, j, q is transmitted by the scheduler i selection bag of input port.After packet leaves input port, crosspoint will converge at this from the packet of different input ports, notice that these bags may have different service requests, for the isolation that guarantees to serve, the input port of packet and output port are regarded as a preface idol, as<i, j 〉, (i ≠ j), each preface couple is answered a buffer queue, is called service quality (QoS) formation, as the Ci among Fig. 6, j, all need all be cached to the corresponding QoS formation from the bag of port j output like this, and the output scheduling device j by crosspoint selects bag to transmit again.
The present invention also provides a kind of method that realizes information exchange, the basic thought of this method is: input port and crosspoint all are provided with buffer queue, input port is divided buffer queue by the destination interface of input packet earlier, further classify according to the demand for services of input packet again, the corresponding buffer queue of each classification, the scheduler of input port selects packet to send to crosspoint according to dispatching algorithm again, crosspoint stores packet in the corresponding cache formation into according to the destination interface of input packet, and the output scheduling device selects packet to transmit to output port according to dispatching algorithm again.
According to the basic thought of said method, in conjunction with Fig. 7 method of the present invention is described again, this method may further comprise the steps:
Step 702~704, input port are provided with a formation for each output port, and the packet of input is cached in the corresponding formation according to its destination interface, further classify the corresponding buffer queue of each classification again according to the demand for services of input packet;
The scheduler of step 705, input port selects packet to send to crosspoint according to dispatching algorithm again, and wherein, what described dispatching algorithm adopted is difference WRR (SPDWRR) algorithm that support is seized;
The present invention also provides a kind of dispatching algorithm of special use, and this basic idea is: buffer queue adopts the chain type storage organization, has high-priority queue Q0 in the buffer queue, and high-priority queue is not participated in the polling sequence queuing of other formation; Scheduler when judging the next formation that obtains dispatching, according to the balance counter C0 of high-priority queue Q0 and the packet length L0 of its team relatively determine whether obtain service, if C0-L0 〉=0, Q0 obtains service, otherwise dispatches the formation in the snoop queue.Individual queue is equipped with balance counter Ci, but the counting mode of the counter C0 of formation Q0 is different with other formation.
Described dispatching algorithm comprises the scheduling of high Priority Queues and the scheduling of Low Priority Queuing.With reference to Fig. 8, Fig. 8 is the high-priority queue scheduling flow figure of dispatching algorithm of the present invention, and described high-priority queue scheduling flow may further comprise the steps:
Step 801~803, scheduler visit high-priority queue judge whether described formation is empty, if, C0=0 is set, process ends, wherein, C0 is the balance counter of high-priority queue, otherwise execution in step 804;
Whether step 805~807, the balance counter C0 that judges a high-priority queue and team packet length L0 difference is more than or equal to 0, if team's head bag of schedules high priority formation is exported, after scheduling is finished, C0=C0-L0, up to C0-L0<0, otherwise process ends.
With reference to Fig. 9, Fig. 9 is the Low Priority Queuing scheduling flow figure of dispatching algorithm of the present invention, and described Low Priority Queuing scheduling flow may further comprise the steps:
Step 901~903, scheduler visit Low Priority Queuing judge whether described formation is empty, if Ci=0 is set, process ends.Wherein, Ci (i=1,2 ..., n-1, n are the formation sum) and be the balance counter of Low Priority Queuing, otherwise execution in step 904;
Whether the difference of step 905~907, the balance counter Ci that judges Low Priority Queuing and the packet length Li of team more than or equal to 0, if, dispatch team's head bag output of described formation, after scheduling is finished, Ci=Ci-Li, otherwise process ends;
The scheduling flow of comprehensive above-mentioned high-priority queue and Low Priority Queuing, described dispatching algorithm comprises following basic step:
A, input packet are divided into the service queue of some, every kind of corresponding buffer queue of service, and in buffer queue, have highest-priority queue Q0, Q0 does not participate in the polling sequence queuing of other formations.Wherein, the chain type storage organization is all adopted in formation, be called linked queue, as shown in figure 10, be that each formation is made up of a single linked list, and for single linked list is provided with head pointer and tail pointer, this is for the ease of the insertion of node and deletion, a balance counter Ci who is used for the memory word joint number is all safeguarded in each formation, and the array of pointers that definition is made up of n pointer is safeguarded this n formation.
B, according to the grade of service and output link total bandwidth, calculate the transmission byte count Ni that each formation should distribute when each poll, this numerical value leave in Ci in the balance counter of each formation (i=1,2 ..., n-1; N is the formation sum); Scheduler is each non-empty queue of poll successively, all Ci=0 of initial time, and when being polled to formation Qi at every turn, Ci=Ci+Ni.
Whether c, the balance counter C0 that judges a highest-priority queue Q0 and team packet length L0 difference more than or equal to 0, if, team's head bag output of scheduling Q0, after scheduling is finished, C0=C0-L0, up to C0-L0<0, execution in step d again; If not, also execution in step d;
Whether the difference of d, the balance counter Ci that judges snoop queue Qi and the packet length Li of team is more than or equal to 0, if team's head bag of scheduling queue Qi is exported, after scheduling is finished, Ci=Ci-Li changes and remove execution in step c, otherwise repeated execution of steps d is until Ci-Li<0.
In addition, the counting mode of the counter C0 of the highest-priority queue Q0 of described dispatching algorithm is different with other formation.
The counting mode of the counter C0 of Q0 is:
I. when Q0 be empty, and scheduler makes C0=C0+N0 when the formation of dispatch service except that Q0, wherein N0 is illustrated in the byte number that a dispatching cycle, interior formation Q0 should get;
Ii. when Q0 obtained scheduled transmission, relatively counter C0 and the packet length L0 of team as if C0-L0 〉=0, then dispatched team's head bag output, and after scheduling was finished, C0=C0-L0 was up to C0-L0<0;
Iii. when Q0 is sky, C0=0 is set.
The balance counter counting mode of all the other formations is identical with former DWRR algorithm, but in the process of these formations of scheduling, supports the gentleness of formation Q0 to seize.All the other formation Qi (the counter mode of 0<i≤n-1) is:
I. when being polled to non-empty queue, counter just obtains byte count Ci=Ci+Ni (0<i≤n-1);
Ii. the length L i of counter Ci and team's head bag relatively, if Ci-Li 〉=0, team's head bag of scheduling queue Qi then, dispatch finish after, Ci=Ci-Li;
Iii. when serve current team head bag, change and go to check whether formation Q0 is empty, if Q0 be sky and C0-L0 〉=0, then Q0 obtains serving, otherwise the operation of repetition ii is up to Ci-Li<0,, then keep the Ci value constant and forward next formation to as if Ci-Li<0;
Iv. when Qi is sky, make Ci=0.Wherein, zero clearing byte count when formation Qi is empty queue, this way are in order not allow Ci infinitely increase, and can prevent the formation that happens suddenly and to the influence of other formations.
By the execution mode of above-mentioned dispatching algorithm as can be known, present embodiment only is preferred implementation of the present invention.Buffer queue can also adopt the sequential storage mode except that can adopting the chain type storage mode.In addition, the chain type storage organization also comprises double linked list, circulation single linked list or circulation double linked list except that single linked list.
More than a kind of system and method for realizing information exchange provided by the present invention and special-purpose dispatching algorithm are described in detail, used specific case herein principle of the present invention and execution mode are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, the part that all can change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (12)
1, a kind of system that realizes information exchange is characterized in that, this system comprises input port and crosspoint, and described input port is used to import packet, and described crosspoint is used to realize the exchanges data of input port and output port;
At input port buffer queue is set, input port is divided each buffer queue, selects packet to be forwarded to crosspoint again, and crosspoint storage packet selects packet to send to output port again.
2, the system of realization information exchange as claimed in claim 1 is characterized in that described buffer queue adopts the VOQ structure.
3, the system of realization information exchange as claimed in claim 1, it is characterized in that, described input port comprises scheduler, is used to judge dispatch service, the packet that described scheduler sends to crosspoint according to the difference WRR SPDWRR algorithm selection of supporting to seize.
4, the system of realization information exchange as claimed in claim 1, it is characterized in that, described crosspoint comprises output buffers formation and output scheduling device, described output buffers formation is used to store the packet that outputs to same destination interface, the described output scheduling device packet that selection sends to output port according to the SPDWRR algorithm.
5, a kind of method that realizes information exchange is characterized in that, this method may further comprise the steps:
A, input port inside are provided with buffer queue, and input port is divided each buffer queue;
B, input port select packet to send to crosspoint;
C, crosspoint storage packet select packet to send to output port again.
6, the method for realization information exchange as claimed in claim 5 is characterized in that, input port is divided specifically each buffer queue and comprised in the steps A:
Input port is provided with a formation for each output port, and the packet of input is cached in the corresponding formation according to its destination interface, further classifies the corresponding buffer queue of each service type again according to the demand for services of input packet.
7, the method for realization information exchange as claimed in claim 5 is characterized in that, step B specifically comprises:
After input port was divided buffer queue, scheduler adopted dispatching algorithm to judge dispatch service according to demand for services, and selected packet to send to crosspoint.
8, the method for realization information exchange as claimed in claim 7 is characterized in that, the algorithm that described scheduler adopts is the SPDWRR algorithm.
9, the method for realization information exchange as claimed in claim 5 is characterized in that, step C specifically comprises:
C1, crosspoint inside are provided with buffer queue, and crosspoint stores packet into the corresponding cache formation according to the destination interface of input packet;
C2, output scheduling device adopt dispatching algorithm to judge dispatch service according to demand for services, and select packet to send to output port.
10, the method for realization information exchange as claimed in claim 9 is characterized in that, the algorithm that the device of output scheduling described in the step C2 adopts is the SPDWRR algorithm.
11, a kind of dispatching algorithm is characterized in that, this algorithm comprises:
A, will import data qualification, the corresponding formation of every class has highest-priority queue Q0 in the formation;
Whether b, scheduler judge the difference of the balance counter C0 of high-priority queue Q0 and the packet length L0 of team more than or equal to 0, if Q0 obtains service, repeating step b is up to C0-L0<0, execution in step c again, otherwise direct execution in step c;
Whether the difference of c, the balance counter Ci that judges snoop queue Qi and the packet length Li of team, changes and remove execution in step b, otherwise execution in step c is until Ci-Li<0 if Qi obtains service more than or equal to 0.
12, dispatching algorithm as claimed in claim 11 is characterized in that, the chain type storage organization is adopted in described formation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2006100765066A CN100420241C (en) | 2006-05-08 | 2006-05-08 | Information switching realizing system and method and scheduling algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2006100765066A CN100420241C (en) | 2006-05-08 | 2006-05-08 | Information switching realizing system and method and scheduling algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1866927A true CN1866927A (en) | 2006-11-22 |
CN100420241C CN100420241C (en) | 2008-09-17 |
Family
ID=37425815
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2006100765066A Expired - Fee Related CN100420241C (en) | 2006-05-08 | 2006-05-08 | Information switching realizing system and method and scheduling algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100420241C (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101945054A (en) * | 2010-10-18 | 2011-01-12 | 西南交通大学 | Dispatching algorithm suitable for feedback two-stage exchange structure |
WO2011050542A1 (en) * | 2009-10-31 | 2011-05-05 | 深圳市利德嘉实业有限公司 | Router constructing method supporting hierarchy of video quality |
CN102356609A (en) * | 2011-08-24 | 2012-02-15 | 华为技术有限公司 | Flow control method of switched network and device |
CN102857442A (en) * | 2011-06-30 | 2013-01-02 | 马维尔国际贸易有限公司 | Method, network device, computer program and computer program product for communication queue state |
CN103516621A (en) * | 2012-06-29 | 2014-01-15 | 中兴通讯股份有限公司 | Method and device for queue scheduling |
CN104734873A (en) * | 2013-12-20 | 2015-06-24 | 深圳市国微电子有限公司 | Switching system and management method and system for buffer in switching equipment in switching system |
CN105024943A (en) * | 2014-04-17 | 2015-11-04 | 罗伯特·博世有限公司 | Method for selecting one of several queues |
CN105915468A (en) * | 2016-06-17 | 2016-08-31 | 北京邮电大学 | Service scheduling method and device |
US9749255B2 (en) | 2011-06-30 | 2017-08-29 | Marvell World Trade Ltd. | Method, network device, computer program and computer program product for communication queue state |
CN109684269A (en) * | 2018-12-26 | 2019-04-26 | 成都九芯微科技有限公司 | A kind of PCIE exchange chip kernel and working method |
CN111865838A (en) * | 2020-07-21 | 2020-10-30 | 深圳市风云实业有限公司 | Multichannel data transmission system of signal |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100375466C (en) * | 2002-04-10 | 2008-03-12 | 中兴通讯股份有限公司 | Data packet forwarding controller and method |
US7243351B2 (en) * | 2002-12-17 | 2007-07-10 | International Business Machines Corporation | System and method for task scheduling based upon the classification value and probability |
CN1728698B (en) * | 2004-07-30 | 2010-08-25 | 国家数字交换系统工程技术研究中心 | Parallel structured order preserved flow equilibrium system, and method for dispatching message based on sorted stream |
CN100514933C (en) * | 2005-01-25 | 2009-07-15 | 南开大学 | Communicating network exchanging system and controlling method based on parallel buffer structure |
-
2006
- 2006-05-08 CN CNB2006100765066A patent/CN100420241C/en not_active Expired - Fee Related
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2011050542A1 (en) * | 2009-10-31 | 2011-05-05 | 深圳市利德嘉实业有限公司 | Router constructing method supporting hierarchy of video quality |
CN101945054B (en) * | 2010-10-18 | 2012-09-19 | 西南交通大学 | Dispatching algorithm suitable for feedback two-stage exchange structure |
CN101945054A (en) * | 2010-10-18 | 2011-01-12 | 西南交通大学 | Dispatching algorithm suitable for feedback two-stage exchange structure |
CN102857442B (en) * | 2011-06-30 | 2017-11-21 | 马维尔国际贸易有限公司 | The method of transmission queue's state, network equipment and products thereof |
US9749255B2 (en) | 2011-06-30 | 2017-08-29 | Marvell World Trade Ltd. | Method, network device, computer program and computer program product for communication queue state |
CN102857442A (en) * | 2011-06-30 | 2013-01-02 | 马维尔国际贸易有限公司 | Method, network device, computer program and computer program product for communication queue state |
WO2012119414A1 (en) * | 2011-08-24 | 2012-09-13 | 华为技术有限公司 | Method and device for controlling traffic of switching network |
CN102356609A (en) * | 2011-08-24 | 2012-02-15 | 华为技术有限公司 | Flow control method of switched network and device |
CN103516621A (en) * | 2012-06-29 | 2014-01-15 | 中兴通讯股份有限公司 | Method and device for queue scheduling |
CN103516621B (en) * | 2012-06-29 | 2017-11-17 | 中兴通讯股份有限公司 | The method and apparatus of queue scheduling |
CN104734873A (en) * | 2013-12-20 | 2015-06-24 | 深圳市国微电子有限公司 | Switching system and management method and system for buffer in switching equipment in switching system |
CN104734873B (en) * | 2013-12-20 | 2018-04-13 | 深圳市国微电子有限公司 | Management method, the system of buffer in a kind of exchange system and its switching equipment |
CN105024943A (en) * | 2014-04-17 | 2015-11-04 | 罗伯特·博世有限公司 | Method for selecting one of several queues |
CN105024943B (en) * | 2014-04-17 | 2020-04-14 | 罗伯特·博世有限公司 | Method and storage medium for selecting a waiting queue from which to extract a transmission data segment |
CN105915468A (en) * | 2016-06-17 | 2016-08-31 | 北京邮电大学 | Service scheduling method and device |
CN105915468B (en) * | 2016-06-17 | 2019-03-29 | 北京邮电大学 | A kind of dispatching method and device of business |
CN109684269A (en) * | 2018-12-26 | 2019-04-26 | 成都九芯微科技有限公司 | A kind of PCIE exchange chip kernel and working method |
CN111865838A (en) * | 2020-07-21 | 2020-10-30 | 深圳市风云实业有限公司 | Multichannel data transmission system of signal |
CN111865838B (en) * | 2020-07-21 | 2022-03-08 | 深圳市风云实业有限公司 | Multichannel data transmission system of signal |
Also Published As
Publication number | Publication date |
---|---|
CN100420241C (en) | 2008-09-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1866927A (en) | Information switching realizing system and method and scheduling algorithm | |
CN1201532C (en) | Quick-circulating port dispatcher for high-volume asynchronous transmission mode exchange | |
CN106533982B (en) | The dynamic queue's dispatching device and method borrowed based on bandwidth | |
US7724760B2 (en) | Method and apparatus for bandwidth guarantee and overload protection in a network switch | |
US8023521B2 (en) | Methods and apparatus for differentiated services over a packet-based network | |
CN102387076B (en) | Shaping-combined hierarchical queue scheduling method | |
EP2740245B1 (en) | A scalable packet scheduling policy for vast number of sessions | |
CN101695051A (en) | Queue length balance dispatching method used for buffered Crossbar | |
WO2013173966A1 (en) | Scheduling method, device and system based on three-stage interconnected switched network | |
CN109639596A (en) | A kind of Scheduling of Gateway method for vehicle-mounted CAN-CANFD hybrid network | |
CN1359241A (en) | Distribution type dispatcher for group exchanger and passive optical network | |
Pan et al. | Localized independent packet scheduling for buffered crossbar switches | |
CN1770744A (en) | Method for preserving delay properties and an apparatus for scheduling transmission link capacity between packet switched telecommunications flows | |
CN1859279A (en) | Method for dispatching variable length data packet queue in crossbar switching matrix | |
CN104333516A (en) | Rotation rotation scheduling method for combined virtual output queue and crosspoint queue exchange structure | |
Dong et al. | Throughput analysis of shared-memory crosspoint buffered packet switches | |
US7460544B2 (en) | Flexible mesh structure for hierarchical scheduling | |
CN102594670B (en) | Multiport multi-flow scheduling method, device and equipment | |
CN106487713A (en) | A kind of service quality multiplexing method and device | |
CN100425035C (en) | Switching system and switching method based on length variable packet | |
CN110430146A (en) | Cell recombination method and switching fabric based on CrossBar exchange | |
WO2009029817A1 (en) | Replicating and switching multicast internet packets in routers using crosspoint memory shared by output ports | |
CN1845526A (en) | Packet equity dispatching method and apparatus | |
Yan et al. | Approaching Work-Conserving Scheduling Algorithm for Mixed Unicast and Multicast in Combined Input and Crosspoint Queued Switch | |
CN111756647A (en) | HQoS service transmission method, device and system |
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: 20080917 Termination date: 20160508 |
|
CF01 | Termination of patent right due to non-payment of annual fee |