CN100471176C - Queue dispatch method and equipment - Google Patents
Queue dispatch method and equipment Download PDFInfo
- Publication number
- CN100471176C CN100471176C CNB2006101275505A CN200610127550A CN100471176C CN 100471176 C CN100471176 C CN 100471176C CN B2006101275505 A CNB2006101275505 A CN B2006101275505A CN 200610127550 A CN200610127550 A CN 200610127550A CN 100471176 C CN100471176 C CN 100471176C
- Authority
- CN
- China
- Prior art keywords
- time
- scheduling
- formation
- queue
- sign
- 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
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The method comprises: according to the number of tokens and the current time information of the time wheel, calculating the scheduling time of the queue to be scheduled; sending the symbol of the queue to be scheduled to the corresponding time memory on the time wheel; when the scheduling time arrives, extracting the symbol of queue to be scheduled, and extracting the scheduling data from corresponding queue to send out; wherein, said time wheel is a data storage device having a pointer and composed of multi storage cells labeled with time value; the rotating speed of the time wheel can be adjusted. The invention also reveals a queue scheduling device thereof.
Description
Technical field
The present invention relates to data communication technology field, relate in particular to a kind of method and device that a plurality of formations are dispatched.
Background technology
At present, along with the develop rapidly of computer networking technology and multimedia technology application day by day deeply and popularize, service quality (QoS) performance of network has been proposed higher, the demand of differentiation more.The QoS of network realizes comprising that a large amount of theory and practice, the present invention only relate to QoS queue scheduling mechanism.
Queue scheduling mechanism guarantees most important to QoS, its main purpose is to guarantee for each Business Stream provides certain service quality, such as, bandwidth, time delay and time delay shake, packet loss etc.In brief, the function of queue scheduling is exactly how network trunk node or router select grouping next to be transmitted from one or more formations.
An effective queue scheduling mechanism depends on the dispatching method that it adopts, and the performance index of weighing array dispatching method mainly contain: equitable degree, delay character, the isolating power to malicious stream, link bandwidth utilization rate, complexity or the like.Common queue scheduling algorithm has RR scheduling (Round Robin), weighted round robin (WRR), weighted-fair scheduling (WFQ) and improved difference circulation (MDRR) etc. at present.
Below only introduce and the immediate MDRR dispatching method of array dispatching method provided by the present invention.
Figure 1 shows that MDRR dispatching method realization block diagram.Its processing procedure comprises: the quene state mapping table to a plurality of formations carries out periodic scan, searches the formation that data are arranged, and each bit of wherein said quene state mapping table represents whether corresponding queues has data; Scheduling token to the QoS user under these formations calculates simultaneously; If formation has data, and QoS user's scheduling token is enough under this formation, then the corresponding bit of this formation position in the request matrix is made as effectively, otherwise it is invalid to be set to; The repeating query module is carried out repeating query according to the cycle repeating query pattern of round robin to the request matrix, after finding effectively to ask, just writes in the dispatch request formation operation dispatching; The size of wherein said request matrix is directly related with number of users, because each user may corresponding a plurality of formations.After scheduling was finished, the QoS information management module was safeguarded the token number that dispatched users takes place according to the formation actual schedule length that feeds back.
In addition, in actual applications, no matter adopt which kind of dispatching method, all also can run into a problem: back-pressure.When carrying out the scheduling of a plurality of formations, must be to the bandwidth of each formation configuration reservation, such as, team A's row are bandwidth of 10Mbps, team B's row are bandwidth of 1Gbps, and third formation is bandwidth of 200Mbps or the like, work as the reserved bandwidth sum of all formations so greater than the final outlet bandwidth, and the total flow of real data greater than the outlet total flow the time, will produce the back-pressure problem; Wherein, the outlet total flow is to be subject to the actual physical outlet, can't change by software approach, just must carry out certain improvement to dispatching algorithm so will solve the back-pressure problem, and the real data flow is controlled.
Adopt described MDRR dispatching method, when running into the back-pressure phenomenon, data in the dispatch request formation shown in Figure 1 will produce accumulation, at this moment, the method that MDRR handles the back-pressure problem is very simple, stops exactly the request matrix is carried out repeating query, so just can in the dispatch request formation, not produce new dispatch request, data in the dispatch request formation have reached a splash line, after promptly back-pressure is removed, just restart repeating query.
The deficiencies in the prior art part mainly comprises:
First, because MDRR is the mode that adopts repeating query, carry out queue scheduling, so when number of users and number of queues are a lot of, such as, 256K user or even 1M user, the time in a week of repeating query will be very long, for same user, the time interval minimum of twice it being dispatched also is the time of a round robin cycle, wherein said round robin cycle has determined the packet rate of final port, and the cycle, long more packet rate was low more.Therefore, for both requiring very high data dispatch speed, under the situation that number of users is very big, the MDRR method is powerless simultaneously;
Second, though in MDRR by the calculating to scheduling token, can guarantee for a long time, user bandwidth guaranteed, but because MDRR is based on the scheduling mechanism of repeating query, therefore, the bigger fluctuation that bandwidth at short notice can occur, the shake (jitter) that concrete reflection is exactly a message is bigger, the message of scheduling is very intensive in the time of can having, sometimes Tiao Du message is more sparse, can't accomplish evenly, and this defective is along with the increase meeting of number of users is more serious;
The 3rd, as can be seen from Figure 1, because adopt repeating query mechanism, so no matter whether the user has data, MDRR can remove to scan the QoS information table synchronously, and the bandwidth pressure for the memory of storing the QoS information table is very big like this, along with the increase of port speed and the increase of number of users, bandwidth is also linear to be increased, and finally makes number of users to increase again;
The 4th, in when, in the scheduling back-pressure phenomenon taking place when, because only adopting, MDRR stops the request matrix is carried out repeating query, but still adopt repeating query mechanism that the scheduling request queue is conducted interviews, from repeating query mechanism essence, the bandwidth of repeating query time-division provisioned user approaches to divide equally, rather than shrinks according to the ratio of reserved bandwidth, and the bandwidth ratio that each formation can be obtained is uncontrollable, and this is inequitable for those high bandwidths user.Because each user spends to buy reserved bandwidth, reserved bandwidth is big, the money of spending is also many, its requirement more benefits of just having the right so, therefore, when in the scheduling back-pressure phenomenon taking place, desirable result is that the formation actual schedule bandwidth of wishing the participation scheduling is shunk according to the ratio of reserved bandwidth.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of method and device of queue scheduling, to realize supporting many formations and two-forty scheduling.
For solving the problems of the technologies described above, the objective of the invention is to be achieved through the following technical solutions:
A kind of array dispatching method comprises:
Obtain and treat scheduling queue token quantity information and time wheel current time information;
According to described token quantity information, bandwidth and decision threshold computational token accumulated time, use described token accumulated time and described time wheel current time information to calculate the scheduling time for the treatment of scheduling queue;
According to described scheduling time, the sign for the treatment of scheduling queue is sent into the time wheel go up the memory cell of corresponding time;
When described scheduling time then, time wheel pointed time wheel is gone up the memory cell of corresponding time, takes out from described memory cell and treats the scheduling queue sign, data dispatching goes out team and sends from respective queue;
Wherein, described time wheel is a data storage device that has pointer, is made of a plurality of memory cell with the time value mark, and the corresponding memory cell of each time point adopts the timeticks timing;
Wherein, described time wheel current time information is a time wheel pointer memory cell time corresponding point pointed, and current time information is along with the rotation of time wheel pointer constantly is updated.
In said method, further comprise:
Judge whether the described scheduling queue sign for the treatment of accumulation occurs, if, then according to treating that the scheduling queue sign gathers way, adjustment time wheel pointer rotary speed, if not, then wheel pointer rotating speed is constant the time;
Wherein, gather way soon, then will take turns speed of finger the time and transfer slowly if treat the scheduling queue sign, otherwise, then will take turns the pointer rotating speed time and transfer fast.
In said method, when described formation is modulated spends but when still having data formation to be scheduled, described method further comprises:
The data length of feedback actual schedule formation according to the data length that feeds back, upgrades the token number of described formation.
In said method, when the described scheduling queue for the treatment of is that described method further comprised when the formation that data newly join the team was arranged:
Check that described formation sign whether in the memory cell of time wheel, if not, then calculates the described queue scheduling time, if, then wait described queue scheduling to finish after, calculate the described queue scheduling time again.
The invention provides a kind of queue scheduling device, described device comprises: QoS information is preserved the unit, scheduling time computing unit, the time wheel unit is treated scheduling queue sign retransmission unit, the data dispatch unit;
Described QoS information is preserved the unit, is used to preserve formation QoS information; Wherein, described formation QoS information comprises formation token information and time wheel current time information at least;
Described scheduling time computing unit, be used for preserving formation token information, bandwidth and the decision threshold computational token accumulated time that the unit is preserved according to QoS information, use described token accumulated time and time wheel current time information to calculate to treat scheduling queue scheduling time;
Described time wheel unit is used for the result of calculation according to the scheduling time computing unit, will treat that the scheduling queue sign is kept at the memory cell of corresponding time;
The described scheduling queue sign retransmission unit for the treatment of is treated scheduling time then when described, is used for going up the corresponding stored space from the time wheel and takes out the formation sign, and described formation sign is transmitted to the data dispatch unit;
Described data dispatch unit is used for going out team's transmission according to transmitting the formation sign that comes from the respective queue data dispatching;
Wherein, described time wheel is a data storage device that has pointer, is made of a plurality of memory cell with the time value mark, and the corresponding memory cell of each time point adopts the timeticks timing;
Wherein, described time wheel current time information is a time wheel pointer memory cell time corresponding point pointed, and current time information is along with the rotation of time wheel pointer constantly is updated.
In said apparatus, further comprise:
Back-pressure state information record cell is used for writing down and treats gathering way of scheduling queue sign retransmission unit formation sign;
Time wheel pointer rotational speed regulation unit is used for gathering way adjustment time wheel pointer rotary speed according to described formation sign; If the formation sign gathers way soon, then the time indicator rotary speed is transferred slowly, otherwise, then will take turns the pointer rotary speed time and transfer fast.
In above-mentioned queue scheduling device, further comprise: data dispatch length feedback unit, formation token updating block;
Wherein, described data dispatch length feedback unit, the data length that data dispatch unit actual schedule is gone out feeds back to formation token updating block;
Wherein, described formation token updating block is used for according to the actual schedule data length, upgrades the token information that QoS information is preserved formation described in the unit.
Above technical scheme has following some beneficial effect as can be seen:
The first, because need not wheel, method provided by the present invention adopts repeating query mechanism, therefore when scheduling, can satisfy very high scheduling rates requirement fully, and scheduling rates is not limited by number of queues;
The second, dispatch very smoothly, because the actual schedule time of data comes out according to the token accurate Calculation,, and can not have contingency as repeating query mechanism even flatness is very good at short notice;
The 3rd, visit to the QoS information table, only when having data to join the team, visit once, visit once during dequeuing data, all the other times can not visit, so just the QoS information table can be stored in outside the sheet, make the user QoS information table support of vast capacity become possibility, nor have the problem of memory access bandwidth;
The 4th, when the back-pressure problem occurring, by adopting the method for adjustment time wheel speed, the assurance user bandwidth is shunk according to the ratio of reserved bandwidth; Adjustment process is dynamic, an adaptive process, no external interference, has realized good fairness and sufficient bandwidth usage.
Description of drawings
Fig. 1 MDRR dispatching method is realized block diagram;
Fig. 2 array dispatching method flow chart of the present invention;
Fig. 3 queue scheduling device of the present invention schematic diagram.
Embodiment
The MDRR method that preamble is addressed is after data enter formation, just judges actually from which queue scheduling data according to the state of all formations.Because whether whether formation have data and formation token enough, can't predict before the scheduling, when number of queues is huge, must adopts the mode of repeating query to carry out queue scheduling, thereby bring many defectives of being mentioned in the background technology.
Solve the defective that above-mentioned dispatching method brings is at all, do not adopt repeating query mechanism in the scheduling, therefore, the present invention proposes a kind of distinct method, its core concept is: when data enter formation, just judge whether described formation has data, and whether the token quantity of described formation enough sends data, after judging end, carry out follow-up scheduling actions again.But, carry out above-mentioned judgement when advanceing to data and join the team, can produce a new problem, though be that described formation is current data are arranged, it is not enough to send the required token number of data, so current this formation can not send data at this moment, can not dispatch, must be through after a while, when waiting described formation token number to reach can to send the requiring of data, could allow this queuing data of scheduling to go out team and send.For above-mentioned judged result, must must preserve, if do not preserve, will run into the problem the same with the MDRR method, later scheduling process must adopt repeating query mechanism data dispatching from formation to go out team and send.
Adopt repeating query mechanism in follow-up scheduling, solve many defectives of being brought owing to repeating query mechanism, the present invention adopts the time wheel as storage device, and the scheduling queue sign is treated in storage;
Wherein, described time wheel is a data storage device that has pointer;
For the ease of understanding, use clock and watch understanding time wheel, be divided into different time points on the clock and watch, suppose that these clock and watch have only hour hands, the time difference of adjacent time point will be 60 minutes on the clock and watch so, when timing reached 60 minutes, time indicator will point to the time point that is adjacent by current point in time;
Time wheel storage device of the present invention is made of a plurality of memory cell and the pointer that mark with the time, the corresponding memory cell of each time point, the present invention adopts the timeticks timing, the time wheel is gone up the difference of the corresponding time point of consecutive storage unit, be to may be prescribed as the time interval such as 32 beats or 64 beats etc., concrete root of number is the physical outlet speed decision of border system factually, when timing reaches 32 beats or 64 beats, time wheel pointer will point to the pairing memory cell of time point that is adjacent by the pairing memory cell of current point in time, the constantly rotation in time of time wheel pointer, the time wheel pointer pairing time point of memory cell pointed is called wheel current time time;
Specifically introduce the present invention below in conjunction with accompanying drawing and adopt time wheel dispatching method and device that storage device provided:
With reference to figure 2, the main flow chart for method provided by the present invention comprises:
Step 201: obtain and treat scheduling queue token quantity information and time wheel current time information;
Wherein, one skilled in the art will appreciate that described token information and time information all are kept in the affiliated user QoS information table of formation;
Wherein, described time wheel current time information is meant time wheel pointer memory cell time corresponding point pointed;
Wherein, every change one next state of described time wheel pointer, current time information promptly is updated once;
Step 202:, calculate the scheduling time for the treatment of scheduling queue according to described token quantity information and current time information;
Wherein, described calculating treats that the scheduling time of scheduling queue specifically comprises:
According to described token quantity, calculate described formation token accumulated time; The computational token accumulated time can be used following formula:
Current token number+bandwidth * token accumulated time 〉=decision threshold
Above-mentioned relation formula implication is: because each token only allows formation to send the bit of some, and user's reserved bandwidth has determined the speed that this User Token increases, therefore, have only the affiliated QoS user's of formation token number to reach more than or equal to sending the required minimum token number of current data packet through a period of time accumulation, when being decision threshold, just can carry out follow-up scheduling actions;
Add the current time with calculating the token accumulated time, treat the scheduling time of formation exactly;
Step 203:, the sign for the treatment of scheduling queue is sent into the time wheel go up the memory cell of corresponding time according to described scheduling time;
Step 204: when described scheduling time then, take out from described memory cell and to treat the scheduling queue sign, data dispatching goes out team and sends from respective queue;
Wherein, time wheel pointer along with the time in continuous rotation, so when the clock rhythmic meter during to described scheduling time, the time takes turns pointer and then can point to pairing memory space of described scheduling time, the formation sign is taken out in the indication of wheel pointer according to the time, carries out the follow-up data scheduling;
After queue scheduling finished, described method further comprised:
Step 205: the real data length of feedback queue scheduling, according to described real data length, upgrade the token number of formation;
Wherein, according to feeding back data length, upgrade described formation token information reason and be: the data of the certain byte of every transmission need deletion and send the token number that size of data equates;
Wherein, can not carry out this step for the formation that does not need to dispatch again after the scheduling;
Wherein, treat described in the step 201 that scheduling queue comprises two kinds: the formation that has data newly to join the team, the formation that after at least once dispatching, still has data to send in the formation;
At these two kinds of formations, described dispatching method is slightly different, below describes respectively:
(1) formation that has data newly to join the team
For the formation that has new data to join the team, before calculating its scheduling time, also comprise and check that whether formation sign that data newly the join the team memory cell on the time wheel is arranged; If this formation sign not in the described memory cell then calculates the queue scheduling time according to the preceding method of having addressed; If the sign of this formation is arranged in the described memory cell, after then waiting this queue scheduling to finish, calculate the described queue scheduling time, treat that scheduling queue is through once scheduling this moment, the formation token number of using when wherein calculating is after the renewal of real data scheduling length;
Wherein, checking whether the formation sign that has data newly to join the team has taken turns the pairing memory cell of time point in the time, is exactly to check whether described formation has data and check whether described formation token is enough in fact; If the formation sign is the pairing memory space of time point on the time wheel, then explanation formation this moment has data and token quantity enough, otherwise, illustrate that then current queue token quantity is not enough, need to wait for;
(2) modulatedly spend but still have data formation to be scheduled
Spend but the people has data formation to be scheduled for modulated, calculate the described scheduling queue for the treatment of during the time, after the formation token number of institute's foundation is the preceding finishing scheduling of this formation, real data scheduling length according to feedback, formation token number after the renewal, therefore, further comprise step for the described method of modulated formation of spending: feedback real data scheduling length, upgrade the formation token;
What more than analyze is when carrying out the actual arrays scheduling generally speaking, method provided by the present invention, and when the back-pressure before running into, addressed of concrete analysis below, this phenomenon, method provided by the present invention:
In case the back-pressure phenomenon takes place, the formation sign that is taken out in the time wheel memory cell will produce accumulation, ideal results is that the formation actual schedule bandwidth of wishing the participation scheduling is shunk according to the ratio of reserved bandwidth when solving the back-pressure problem, in order to reach this effect, the preceding method of having addressed provided by the invention further comprises:
Judge whether the described scheduling queue sign for the treatment of accumulation occurs, if, then according to treating that the scheduling queue sign gathers way, adjustment time wheel pointer rotary speed, if not, then wheel pointer rotating speed is constant the time;
Wherein, gather way soon, then will take turns speed of finger the time and transfer slowly if treat the scheduling queue sign, otherwise, then will take turns the pointer rotating speed time and transfer fast
Such as, point to adjacent time point pairing memory space every 32 bat times wheel pointer by the pairing memory space of current point in time by original, be adjusted to every 64 bats, time wheel pointer changes a next state;
Carrying out the reason that such action can arrive beneficial effect noted earlier is:
Because the queue scheduling time is to take turns Time Calculation according to the time of not adjusting the time interval, the formation sign put into the time interval is adjusted on big time wheel, the time that is equivalent to queue scheduling is extended, so corresponding, the bandwidth of the actual acquisition of these formations is also shunk with regard to equal proportion, this shrinkage rates just is exactly the time indicator ratio slack-off with respect to original speed, in addition, because when regulating the time indicator rotating speed, it is the speed regulation that increases according to the formation sign, therefore, which has only have the bandwidth of the formation of data to be affected, the bandwidth contraction also only realizes which does not have the formation of data in this process bandwidth not to be influenced in these formations.That is to say that total band width in physical can be shared by these formations that data are arranged, and can not have bandwidth waste.
After the back-pressure phenomenon is removed, slow if the formation sign gathered way, can also the described time interval be adjusted go back as required according to the method described above, improve gathering way of formation sign;
By above narration as can be known, the process in the described adjustment time interval is the feedback regulation mode of a dynamic self-adapting, does not have the interference in any external world.
With reference to figure 3, dispatching device that formation of the present invention provides is described in detail, described device mainly comprises: QoS information is preserved the unit, scheduling time computing unit, the time wheel unit is treated scheduling queue sign retransmission unit, the data dispatch unit;
Described QoS information is preserved the unit, is used to preserve formation QoS information;
Wherein, described formation QoS information comprises formation token information and time wheel current time information at least;
Described scheduling time computing unit, be used for preserving formation token information and the time wheel current time information of preserving the unit according to QoS information, calculate and treat scheduling queue scheduling time;
Described time wheel unit is used for the result of calculation according to the scheduling time computing unit, will treat that the scheduling queue sign is kept at the memory cell of corresponding time;
The described scheduling sign formation retransmission unit for the treatment of treats that when time wheel pointed is described scheduling time during point, be used for going up the corresponding stored space from the time wheel and take out the formation sign, and just described formation sign is transmitted to the data dispatch unit;
Described data dispatch unit is used for going out team's transmission according to transmitting the formation sign that comes from the respective queue data dispatching;
In actual applications, described dispatching device further comprises: data dispatch length feedback unit, formation token updating block;
Wherein, described data dispatch length feedback unit, the data length that data dispatch unit actual schedule is gone out feeds back to formation token updating block;
Wherein, described formation token updating block is used for according to the actual schedule data length, upgrades the token information that QoS information is preserved formation described in the unit;
More than be device provided by the present invention generally speaking, when the back-pressure phenomenon occurring, described device further comprises:
Back-pressure state information record cell is used for writing down and treats gathering way of scheduling queue sign retransmission unit formation sign;
Time wheel pointer rotational speed regulation unit is used for gathering way adjustment time wheel pointer rotary speed according to described formation sign; If the formation sign gathers way soon, then the time indicator rotary speed is transferred slowly, otherwise, then will take turns the pointer rotary speed time and transfer fast;
In addition, as shown in Figure 3, comprise also among the present invention that the data unit of joining the team is used for obtaining packet from external interface the operation of joining the team; Other parts that do not illustrate in the drawings do not relate to the scope of the invention, are not described in detail here.
More than a kind of array dispatching method provided by the present invention and device 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 (7)
1, a kind of array dispatching method is characterized in that, comprising:
Obtain and treat scheduling queue token quantity information and time wheel current time information;
Use formula " current token number+bandwidth * token accumulated time 〉=decision threshold " to calculate the token accumulated time according to token quantity information, bandwidth and decision threshold, use described token accumulated time to add the scheduling time that the above time wheel current time information obtains treating scheduling queue;
According to described scheduling time, the sign for the treatment of scheduling queue is sent into the time wheel go up the memory cell of corresponding time;
When described scheduling time then, time wheel pointed time wheel is gone up the memory cell of corresponding time, takes out from described memory cell and treats the scheduling queue sign, data dispatching goes out team and sends from respective queue;
Wherein, described time wheel is a data storage device that has pointer, is made of a plurality of memory cell with the time value mark, and the corresponding memory cell of each time point adopts the timeticks timing;
Wherein, described time wheel current time information is a time wheel pointer memory cell time corresponding point pointed, and current time information is along with the rotation of time wheel pointer constantly is updated.
According to the described method of claim 1, it is characterized in that 2, described method further comprises:
Judge whether the described scheduling queue sign for the treatment of accumulation occurs, if, then according to treating that the scheduling queue sign gathers way, adjustment time wheel pointer rotary speed, if not, then wheel pointer rotating speed is constant the time;
Wherein, gather way soon, then will take turns speed of finger the time and transfer slowly if treat the scheduling queue sign, otherwise, then will take turns the pointer rotating speed time and transfer fast.
3, according to claim 1 or 2 described methods, it is characterized in that, when described formation is modulated spends but when still having data formation to be scheduled, described method further comprises:
The data length of feedback actual schedule formation according to the data length that feeds back, upgrades the token number of described formation.
4, according to claim 1 or 2 described methods, it is characterized in that, when the described scheduling queue for the treatment of is that described method further comprised when the formation that data newly join the team was arranged:
Check that described formation sign whether in the memory cell of time wheel, if not, then calculates the described queue scheduling time, if, then wait described queue scheduling to finish after, calculate the described queue scheduling time again.
5, a kind of queue scheduling device is characterized in that, described device comprises: QoS information is preserved the unit, scheduling time computing unit, the time wheel unit is treated scheduling queue sign retransmission unit, the data dispatch unit;
Described QoS information is preserved the unit, is used to preserve formation QoS information; Wherein, described formation QoS information comprises formation token information and time wheel current time information at least;
Described scheduling time computing unit, be used for preserving formation token information, bandwidth and the decision threshold computational token accumulated time that the unit is preserved, use described token accumulated time and described time wheel current time information to calculate to treat scheduling queue scheduling time according to QoS information;
Described time wheel unit is used for the result of calculation according to the scheduling time computing unit, will treat that the scheduling queue sign is kept at the memory cell of corresponding time;
The described scheduling queue sign retransmission unit for the treatment of is treated scheduling time then when described, is used for going up the corresponding stored space from the time wheel and takes out the formation sign, and described formation sign is transmitted to the data dispatch unit;
Described data dispatch unit is used for going out team's transmission according to transmitting the formation sign that comes from the respective queue data dispatching;
Wherein, described time wheel is a data storage device that has pointer, is made of a plurality of memory cell with the time value mark, and the corresponding memory cell of each time point adopts the timeticks timing;
Wherein, described time wheel current time information is a time wheel pointer memory cell time corresponding point pointed, and current time information is along with the rotation of time wheel pointer constantly is updated.
According to the described device of claim 5, it is characterized in that 6, described device further comprises:
Back-pressure state information record cell is used for writing down and treats gathering way of scheduling queue sign retransmission unit formation sign;
Time wheel pointer rotational speed regulation unit is used for gathering way adjustment time wheel pointer rotary speed according to described formation sign; If the formation sign gathers way soon, then the time indicator rotary speed is transferred slowly, otherwise, then will take turns the pointer rotary speed time and transfer fast.
7, according to claim 5 or 6 described devices, it is characterized in that described dispatching device further comprises: data dispatch length feedback unit, formation token updating block;
Wherein, described data dispatch length feedback unit, the data length that data dispatch unit actual schedule is gone out feeds back to formation token updating block;
Wherein, described formation token updating block is used for according to the actual schedule data length, upgrades the token information that QoS information is preserved formation described in the unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006101275505A CN100471176C (en) | 2006-09-12 | 2006-09-12 | Queue dispatch method and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006101275505A CN100471176C (en) | 2006-09-12 | 2006-09-12 | Queue dispatch method and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1960334A CN1960334A (en) | 2007-05-09 |
| CN100471176C true CN100471176C (en) | 2009-03-18 |
Family
ID=38071832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006101275505A Expired - Fee Related CN100471176C (en) | 2006-09-12 | 2006-09-12 | Queue dispatch method and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100471176C (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101146091B (en) * | 2007-09-05 | 2010-09-08 | 中兴通讯股份有限公司 | Multi-channel data output method and system |
| CN102455932B (en) * | 2010-10-22 | 2014-06-25 | 金蝶软件（中国）有限公司 | Serial execution method, device and system for task instances |
| CN101969409B (en) * | 2010-11-05 | 2013-05-08 | 南京邮电大学 | Queue scheduling method based on timer and MDRR |
| CN103229465A (en) * | 2012-11-12 | 2013-07-31 | 华为技术有限公司 | Message sending method and device |
| CN105637475B (en) * | 2014-09-16 | 2019-08-20 | 华为技术有限公司 | Concurrent access method and system |
| CN108259355B (en) | 2014-12-30 | 2022-03-11 | 华为技术有限公司 | Message forwarding method and device |
| CN107870820B (en) * | 2016-09-23 | 2023-10-13 | 北京忆恒创源科技股份有限公司 | Providing quality of service in solid state drives |
| CN108446777B (en) * | 2017-02-16 | 2022-04-29 | 菜鸟智能物流控股有限公司 | Storage space management method and related equipment |
| CN108063734A (en) * | 2017-12-05 | 2018-05-22 | 郑州云海信息技术有限公司 | A kind of network resource scheduling method and device |
| CN111290786B (en) * | 2018-12-12 | 2022-05-06 | 展讯通信（上海）有限公司 | Information processing method, device and storage medium |
| CN112364067B (en) * | 2021-01-12 | 2022-02-11 | 绿城科技产业服务集团有限公司 | Large-batch data exporting system and exporting method thereof |
| CN114326560B (en) * | 2021-11-18 | 2024-02-09 | 北京华能新锐控制技术有限公司 | Method and device for reducing CPU load of domestic PLC of wind turbine generator |
-
2006
- 2006-09-12 CN CNB2006101275505A patent/CN100471176C/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN1960334A (en) | 2007-05-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100471176C (en) | Queue dispatch method and equipment | |
| Meng et al. | Online deadline-aware task dispatching and scheduling in edge computing | |
| Shah-Mansouri et al. | Joint optimal pricing and task scheduling in mobile cloud computing systems | |
| Zhang et al. | Moving big data to the cloud: An online cost-minimizing approach | |
| US8730982B2 (en) | Scheduling of data transmission with minimum and maximum shaping of flows in a network device | |
| Ramakrishnan | Performance considerations in designing network interfaces | |
| US6122673A (en) | Port scheduler and method for scheduling service providing guarantees, hierarchical rate limiting with/without overbooking capability | |
| CN104572307B (en) | The method that a kind of pair of virtual resource carries out flexible scheduling | |
| US7929438B2 (en) | Scheduler pipeline design for hierarchical link sharing | |
| CN103229145B (en) | Lock-less buffer management scheme for communication network application | |
| CN102387076B (en) | Shaping-combined hierarchical queue scheduling method | |
| CN103561354B (en) | Method and device for calculating and processing video smoothness | |
| CN103294521A (en) | Method for reducing communication loads and energy consumption of data center | |
| CN1495588A (en) | Management method of network interface module energy consumption in wireless computing equipment and its system | |
| FR2578071A1 (en) | MULTITRAITE INSTALLATION WITH SEVERAL PROCESSES | |
| CN104410870B (en) | Method for distributing bandwidth resource of streaming media application with optimized energy consumption | |
| Herrería-Alonso et al. | How efficient is energy-efficient ethernet? | |
| CN102882809B (en) | Network speed-limiting method and device based on message buffering | |
| EP1491021A1 (en) | Rate-based scheduling for packet applications | |
| Li et al. | Co-Scheduler: A coflow-aware data-parallel job scheduler in hybrid electrical/optical datacenter networks | |
| Zhang et al. | ATFQ: a fair and efficient packet scheduling method in multi-resource environments | |
| CN103684866B (en) | A kind of distribution method of dynamic bandwidth of ethernet passive optical network | |
| Gong et al. | Maximizing rewards in wireless networks with energy and timing constraints for periodic data streams | |
| Li et al. | Deep reinforcement learning-based power and caching joint optimal allocation over mobile edge computing | |
| CN115567466A (en) | Time-sensitive traffic online scheduling method and device based on deep reinforcement learning |
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: 20090318 Termination date: 20150912 |
|
| EXPY | Termination of patent right or utility model |