CN101848149B - Method and device for scheduling graded queues in packet network - Google Patents
Method and device for scheduling graded queues in packet network Download PDFInfo
- Publication number
- CN101848149B CN101848149B CN2010101585993A CN201010158599A CN101848149B CN 101848149 B CN101848149 B CN 101848149B CN 2010101585993 A CN2010101585993 A CN 2010101585993A CN 201010158599 A CN201010158599 A CN 201010158599A CN 101848149 B CN101848149 B CN 101848149B
- Authority
- CN
- China
- Prior art keywords
- formation
- node
- group
- grouping
- scheduling
- 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
Abstract
The invention provides a method and a device for scheduling graded queues in a packet network, wherein the method comprises the steps of: progressively grading the queues according to network performance requirements of each queue, and forming a priority queue group of a tree structure; extracting grouped category identifiers from received grouped packet headers, and classifying the groups into aqueue tail according to the category identifiers; and scheduling the queues in the current priority queue group by selecting a strict priority queue scheduling method or a probability-based queue scheduling method according to the priority queue group of the tree structure. The device comprises a queue grading module, a grouping and classifying module and a queue scheduling module. The embodimentrealizes enhancing the network performance of a low-priority queue while strictly ensuring the network performance of a high-priority queue.
Description
Technical field
The present invention relates to field of computer technology, relate in particular to a kind of method and apparatus of scheduling graded queues in packet network.
Background technology
Along with the rapid growth of miscellaneous service in the network, people more and more are concerned about the service quality of network, and queue scheduling algorithm has caused network research person's extensive concern in recent years as a key technology that guarantees network efficiency and practicality.
Queue scheduling algorithm of the prior art comprises strict priority (Strict Priority; Hereinafter to be referred as: the SP) queue scheduling algorithm of queue scheduling algorithm and Based on Probability.Wherein, the SP queue scheduling algorithm is that grouping is classified, classification according to grouping is sent grouping into corresponding formation, the grouping that belongs to higher priority queues after the classification is preferentially sent, and the grouping that belongs to the lower priority formation is then waited for just robbing when the grouping without higher priority sends and accounted for channel.The queue scheduling algorithm of Based on Probability then is probability of each queue assignment, when each queue scheduling, begins poll from first formation, when the formation non-NULL that is polled to, then dispatches with the probability that distributes.
In realizing process of the present invention, the inventor finds that there are the following problems at least in the prior art: the SP queue scheduling algorithm can guarantee the absolute priority of each formation, but to sacrifice lower priority network of queues performance as cost, the lower priority formation may can not be scheduled for a long time when network is busy for the assurance of higher priority queues network performance for it.Although the queue scheduling algorithm of Based on Probability has certain assurance to the network performance of lower priority formation, cut down the strict guarantee to the network performance of higher priority queues.Therefore, these two kinds of methods can't be taken into account the network performance of higher priority queues and lower priority formation.
Summary of the invention
The invention provides a kind of method and apparatus of scheduling graded queues in packet network, in order to solve the defective of the network performance that can't take into account higher priority queues and lower priority formation in the prior art, realization improves the network performance of Low Priority Queuing in the network performance of strict guarantee high-priority queue.
The invention provides a kind of method of scheduling graded queues in packet network, comprising:
Performance index requirements according to individual queue is divided step by step to described formation, and forms priority query's group of tree structure;
From the packets headers of the grouping that receives, extract the classification logotype of described grouping, the formation afterbody under according to described classification logotype described grouping being included into;
According to priority query's group of described tree structure, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched;
Described performance index requirements according to individual queue is divided step by step to described formation, and priority query's group of formation tree structure specifically comprises:
First according to the scope of determining the network performance value in all formations from minimum network performance value to the highest network performance value, and this network performance value scope is divided into a series of little range sections, the network performance value is close and formation that belong to same range section is classified as same priority query group.The rest may be inferred, step by step the prioritization set of queues.The priority query's group that to the last once is divided into satisfies actual requirement.Then the priority query's group that represents the maximum at all formation places with a root node, root node is father node, the several priority queries group that is divided into for the first time is expressed as corresponding child node, the child node of dividing for the first time is father node, and the priority query's group that is divided into for the second time is expressed as the child node of corresponding father node.The like, until all formations all are illustrated in the tree structure with the form of the leaf node of reality;
Described priority query's group according to described tree structure selects the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched, and specifically comprises:
Search the child node of every one-level of tree structure and the subordinate relation of even higher level of node, if be subordinated to the child node number of same even higher level of node greater than 1, and described child node is physical node, then adopts the queue scheduling method of Based on Probability that the formation in current priority query's group is dispatched; If be subordinated to the child node number of same even higher level of node greater than 1, and described child node is not physical node, then adopts the Priority Queuing method that the formation in current priority query's group is dispatched; Wherein, described physical node is leaf node;
The father node that child node corresponding to the formation for the treatment of output grouping of obtaining is replaced described child node, every one-level from top to bottom to described tree structure repeats above-mentioned steps successively, replaced until the root node of described tree structure finally treats child node corresponding to the formation of output grouping, and from the formation of the child node representative that replaces described root node output grouping.
The invention provides a kind of device of scheduling graded queues in packet network, comprising:
Module is divided in formation, is used for according to the performance index requirements of individual queue described formation being divided step by step, and forms priority query's group of tree structure;
Grouping and classifying module is used for extracting the classification logotype of described grouping, the formation afterbody under according to described classification logotype described grouping being included into from the packets headers of the grouping that receives;
The queue scheduling module is used for the priority query's group according to described tree structure, selects the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched;
Module is divided in described formation, the concrete scope that is used for determining to the highest network performance value from minimum network performance value according to all formations first the network performance value, and this network performance value scope is divided into a series of little range sections, the network performance value is close and formation that belong to same range section is classified as same priority query group; The rest may be inferred, step by step the prioritization set of queues; The priority query's group that to the last once is divided into satisfies actual requirement; Then the priority query's group that represents the maximum at all formation places with a root node, root node is father node, the several priority queries group that is divided into for the first time is expressed as corresponding child node, the child node of dividing for the first time is father node, and the priority query's group that is divided into for the second time is expressed as the child node of corresponding father node; The like, until all formations all are illustrated in the tree structure with the form of the leaf node of reality;
Described queue scheduling module specifically comprises:
Search the unit, for the child node of every one-level of searching tree structure and the subordinate relation of even higher level of node;
The first scheduling unit, be used for when the child node number that be subordinated to same even higher level of node greater than 1, and described child node is physical node, then adopts the queue scheduling method of Based on Probability that the formation in current priority query's group is dispatched;
The second scheduling unit, be used for when the son node number that be subordinated to same even higher level of node according to greater than 1, and described child node is not physical node, then adopts the Priority Queuing method that the formation in current priority query's group is dispatched; Wherein, described physical node is leaf node;
The output acquiring unit, child node corresponding to the formation for the treatment of output grouping that is used for obtaining replaces the father node of described child node, every one-level from top to bottom to described tree structure repeats above-mentioned steps successively, replaced until the root node of described tree structure finally treats child node corresponding to the formation of output grouping, and from the formation of the child node representative that replaces described root node output grouping.
The method and apparatus of scheduling graded queues in packet network of the present invention, by according to the performance index requirements of individual queue in the network formation being divided step by step, and priority query's group of formation tree structure, classification logotype according to the grouping that receives is included into affiliated formation afterbody with grouping, then according to priority query's group of tree structure, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in current priority query's group is dispatched; The present embodiment has solved the defective that can't take into account the network performance of higher priority queues and lower priority formation in the prior art, has realized having improved the network performance of Low Priority Queuing in the network performance of strict guarantee high-priority queue.
Description of drawings
In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art, the below will do one to the accompanying drawing of required use in embodiment or the description of the Prior Art and introduce simply, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.
Fig. 1 is the flow chart of the embodiment of the method one of scheduling graded queues in packet network of the present invention;
Fig. 2 is the flow chart of the embodiment of the method two of scheduling graded queues in packet network of the present invention;
Fig. 3 is the tree structure figure of the embodiment of the method two medium priority set of queues of scheduling graded queues in packet network of the present invention;
Fig. 4 is the structural representation of the device embodiment one of scheduling graded queues in packet network of the present invention;
Fig. 5 is the structural representation of the device embodiment two of scheduling graded queues in packet network of the present invention.
Embodiment
For the purpose, technical scheme and the advantage that make the embodiment of the invention clearer, below in conjunction with the accompanying drawing in the embodiment of the invention, technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that obtains under the creative work prerequisite.
Fig. 1 is the flow chart of the embodiment of the method one of scheduling graded queues in packet network of the present invention, and as shown in Figure 1, the present embodiment provides a kind of method of scheduling graded queues in packet network, can comprise the steps:
The present embodiment provides a kind of method of scheduling graded queues in packet network, by according to the performance index requirements of individual queue in the network formation being divided step by step, and priority query's group of formation tree structure, classification logotype according to the grouping that receives is included into affiliated formation afterbody with grouping, then according to priority query's group of tree structure, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in current priority query's group is dispatched; The present embodiment has solved the defective that can't take into account the network performance of higher priority queues and lower priority formation in the prior art, has realized having improved the network performance of Low Priority Queuing in the network performance of strict guarantee high-priority queue.
Fig. 2 is the flow chart of the embodiment of the method two of scheduling graded queues in packet network of the present invention, and as shown in Figure 2, the present embodiment provides a kind of method of scheduling graded queues in packet network, can comprise the steps:
Step 201, the packet class in the statistical packet network and every kind of performance index requirements that grouping is corresponding, and in buffering area, carry out buffer memory for every kind of grouping arranges independently equal-sized formation.
In packet network, because the variation of the business that the user uses, so that the grouping kind in the packet network is also corresponding varied, such as the grouping of audio frequency and video class, the grouping of image class etc.Before the technical scheme of carrying out the present embodiment, this step is by existing all categories of grouping to add up to network internal, obtain the classification of each grouping in the packet network, and add up every kind the grouping corresponding performance index requirements, network performance herein can comprise time delay, packet loss, throughput and delay variation etc., and the corresponding performance index requirements value of concrete employing represents in the present embodiment.Simultaneously, open up independently equal and opposite in direction for every kind of grouping counting on and buffer memory is carried out in certain formation in buffering area, the packet size in each formation is fixing and equates.In the present embodiment, suppose to have in the network 10 kinds of groupings, then be respectively these 10 kinds of groupings and open up 10 buffer queues in buffering area, network performance describes as an example of time delay example in the present embodiment, and the delay performance for the individual queue that counts on as shown in table 1 below requires:
The delay performance of table 1 individual queue requires statistics
| The formation numbering | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| Delay performance requirement/ms | 29 | 15 | 36 | 22 | 45 | 90 | 21 | 80 | 10 | 55 |
Before carrying out queue scheduling, according to the performance index requirements of individual queue formation is divided step by step first, and form priority query's group of tree structure.Namely according to the difference of formation for performance index requirements, prioritization set of queues step by step, the priority query's group that to the last once is divided into satisfies actual requirement.And the form of ready-portioned priority query group with node be illustrated in the tree structure, until all formations all are illustrated in the tree structure with the form of leaf node.Particularly, first according to the scope of determining the network performance value in all formations from minimum network performance value to the highest network performance value, and this network performance value scope is divided into a series of little range sections, the network performance value is close and formation that belong to same range section is classified as same priority query group.The rest may be inferred, step by step the prioritization set of queues.The priority query's group that to the last once is divided into satisfies actual requirement.Then the priority query's group that represents the maximum at all formation places with a root node, root node is father node, the several priority queries group that is divided into for the first time is expressed as corresponding child node, the child node of dividing for the first time is father node, and the priority query's group that is divided into for the second time is expressed as the child node of corresponding father node.The like, until all formations all are illustrated in the tree structure with the form of the leaf node of reality.
Still describe as an example of above-mentioned table 1 example, when in the present embodiment formation being divided, require as dividing foundation take its corresponding delay performance, wherein, minimum requirements value to time delay is 10ms, and the peak demand value is 90ms, i.e. the scope of delay performance requirement is 10ms-90ms, is assumed to be the A0 group in the present embodiment.Fig. 3 is the tree structure figure of the dispatching method embodiment two medium priority set of queues of graded queues in the packet network of the present invention, and then the first order in the tree structure is root node 0, and namely the A0 corresponding to priority query's group organizes.
The scope that can first delay performance be required is divided into two less range sections, be 10ms-50ms, 51ms-90ms, delay performance required value according to individual queue in the above-mentioned table 1, priority query's group of dividing this moment can be assumed to be A11 group (10ms-50ms) and A12 group (51ms-90ms), wherein, the A11 group comprises formation 9,2,7,4,1,3,5, and the A12 group comprises formation 10,8,6.At this moment, formation is divided into two-stage, and the second level node 11 and 12 in the tree structure represents respectively A11 and A12.
According to above-mentioned classification division methods, continuation is divided the group A11 of priority query and the A12 after dividing, for A11, individual queue is 10ms to the minimum requirements value of time delay, and the peak demand value is 50ms, the scope that delay performance requires can be divided into two less range sections, be 10-30ms, 31-45ms, delay performance required value according to individual queue in the above-mentioned table 1, priority query's group of dividing this moment can be assumed to be A14 group (10ms-30ms) and A13 group (31ms-45ms), wherein, the A14 group comprises formation 9,2,7,4,1, A13 group comprises formation 3,5.For A12, individual queue is 55ms to the minimum requirements value of time delay, and the peak demand value is 90ms, the scope that delay performance can be required be divided into two less range sections, i.e. 55-70ms, 71-90ms are according to the delay performance required value of individual queue in the above-mentioned table 1, priority query's group of dividing this moment can be assumed to be A15 group (55ms-70ms) and A16 group (71ms-90ms), wherein, the A15 group comprises formation 8,6, and the A16 group comprises formation 10.At this moment, formation is divided into three grades, and the third level node 13 in the tree structure, 14 and 15 represents respectively A13, A14 and A15, because the A16 group includes only formation 10, then it directly represents with node 10.
After above-mentioned two-stage division, because the A16 group only comprises a formation 10, the A13 group only comprises two formations 3 and 5, A15 group only comprises two formations 8 and 6, then no longer above-mentioned three groups is divided.And the A14 group still comprises 9,2,7,4,1 five formations, then according to above-mentioned classification division methods, continuation is divided A14, the scope that delay performance requires can be divided into two less range sections, be 10-20ms, 21-30ms, delay performance required value according to individual queue in the above-mentioned table 1, priority query's group of dividing this moment can be assumed to be A17 group (10ms-20ms) and A18 group (21ms-30ms), wherein, the A17 group comprises formation 9,2, and the A18 group comprises formation 7,4,1.At this moment, formation is divided into level Four, and the fourth stage node 16 and 17 in the tree structure represents respectively A18 and A17.Because the A17 group comprises formation 9,2, the A18 group comprises formation 7,4,1 then with the node 9 in the tree structure and 2 child nodes as node 17, with node 7,4,1 child node as node 16, then can obtain Pyatyi tree structure as shown in Figure 3.
In the present embodiment, can represent leaf node with physical node, represent that with empty node non-leaf node represents, namely physical node can represent priority query's group that the number of queues that comprises equals 1, and empty node can represent that the number of queues that comprises is greater than priority query's group of 1.
When having grouping to transmit in the packet network, then from the packets headers of the grouping that receives, extract the classification logotype of this grouping, identify the formation of determining that this grouping is affiliated according to this classification.Particularly, this step can be obtained the affiliated formation of this grouping for the first classification logotype according to grouping, and judges the length the higher limit whether length of this formation equals to preset, if, show that then this formation is full, then this grouping is abandoned, it is not carried out follow-up transmission process; If the length of the formation under should grouping less than default length higher limit, then show this formation less than, the formation afterbody under then this grouping being included into is put into affiliated rear of queue with this grouping.
When carrying out queue scheduling, the present embodiment can according to priority query's group of the tree structure of above-mentioned formation, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in current priority query's group is dispatched.Particularly, this step 204 can specifically comprise the steps: at first, search the child node of every one-level of tree structure and the subordinate relation of even higher level of node, if be subordinated to the child node number of same even higher level of node greater than 1, and described child node is physical node, then adopts the queue scheduling method of Based on Probability that the formation in current priority query's group is dispatched; If the son node number that is subordinated to same even higher level of node is according to greater than 1, and described child node is not physical node, then adopts the Priority Queuing method that the formation in current priority query's group is dispatched; Wherein, described physical node is leaf node.Secondly, the father node that child node corresponding to the formation for the treatment of output grouping of obtaining is replaced described child node, every one-level from top to bottom to described tree structure repeats above-mentioned steps successively, replaced until the root node of described tree structure treats child node corresponding to the formation of output grouping, and from the formation of the child node representative that replaces described root node output grouping.
Still describe as an example of priority query's group of forming tree structure as shown in Figure 3 example herein, when carrying out this step, suppose that working as prime is the superlative degree of tree structure, present node is first node in the next stage of prime, namely working as prime is level V, and present node is the leaf node 3 in the fourth stage.Searching present node 3 does not have child node in level V, then skip this node 3, in like manner skips node 5 yet, and present node is node 16.Search the child node of present node 16 in level V and have 7,4,1, and be leaf node, then adopt the queue scheduling method of Based on Probability that the formation in priority query's group of node 16 correspondences is dispatched, namely the formation of child node 7,4,1 correspondence is dispatched.When dispatching, first the formation of child node 7,4,1 correspondence detected, if two or more formations are arranged simultaneously not for empty, then dispatched with corresponding probability, select a formation for the treatment of output grouping, namely select a formation that is about to be scheduled; If only have a formation for empty, then directly with this formation as the formation for the treatment of output grouping, namely directly with this formation as the formation that is about to be scheduled.
Particularly, when the queue scheduling method that adopts Based on Probability is dispatched, can be first according to the block length in the formation, and adopt following relational expression to set the scheduling probability of described formation:
Lm∶Ln＝Pn∶Pm (1)
Wherein, Lm and Ln are respectively the block length among any two formation Qm and the Qn, and Pm and Pn are respectively the scheduling probability of described formation Qm and Qn.Scheduling probability and the random random number that generates according to formation are obtained the formation for the treatment of output grouping in current priority query's group, particularly, can adopt following method to carry out queue scheduling.The ratio of supposing the scheduling probability of individual queue Q1, Q2...Qn is p
1: p
2: p
3: p
4: L p
N-1: p
n, then generate according to this ratio and obtain first group of definite value, that is:
...，
Generate and obtain second group of definite value, that is: y according to above-mentioned first group of definite value again
1=x
1, y
2=y
1+ x
2, y
3=y
2+ x
3, y
4=y
3+ x
4..., y
N-1=y
N-2+ x
N-1Then produce a random number R, obtain the formation for the treatment of output grouping in current priority query's group according to the relation of random number R and above-mentioned second group of definite value.If 0＜R＜y
1, the formation for the treatment of output grouping of then obtaining this scheduling is Q1; If y
1＜R＜y
2, the formation for the treatment of output grouping of then obtaining this scheduling is Q2; If y
2＜R＜y
3, the formation for the treatment of output grouping of then obtaining this scheduling is Q3; If ... y
N-1＜R＜1, the formation for the treatment of output grouping of then obtaining this scheduling is Qn.Adopt the queue scheduling method of Based on Probability that child node 7,4, formation Q7, the Q4 of 1 correspondence, Q1 are dispatched herein, suppose that wherein block length L7, L4, L1 are respectively 30,20,50, formation Q7, the Q4 that then arranges as can be known according to above-mentioned formula (1), the scheduling probability P 7 of Q1, P4, P1 satisfied 5: 2: 3.Then the ratio according to above-mentioned scheduling probability obtains and generates first group of definite value: 1/2,1/5 and 3/10, and the second group of definite value that generates according to first group of definite value is: 1/2,7/10.Then in scheduling process, produce one 0 to 1 random number R, suppose that random number R satisfies: 0＜R＜1/2, the formation of then treating output grouping is 7, the formation that is about to be scheduled of namely selecting is 7, if random number R satisfies: 1/2＜R＜7/10, the formation of then treating output grouping is 4, the formation that is about to be scheduled of namely selecting is 4, if random number R satisfies: 7/10＜R＜1, treat that then the formation of output grouping is 1, the formation that is about to be scheduled of namely selecting is 1.Suppose that the formation for the treatment of output grouping of selecting after the process scheduling is 4, then with the formation 4 temporary transient nodes 16 that replace, namely with formation 4 temporary transient formations as present node 16 correspondences.
After node 16 processed, present node is moved to node 17, in like manner selecting the formation for the treatment of output grouping according to said method is 9, and with formation 9 temporarily as the formations of node 17 correspondences.Because node 8 and 6 does not all have child node, then skips node 8 and 6, finishes the scheduling of level V.Work as prime this moment and subtract 1, namely the fourth stage is processed, and present node is moved to first node of the third level, namely node 13.
Repeat said process and carry out the scheduling of the fourth stage, because the child node of present node 13 in the fourth stage is 3,5, and is leaf node, then adopt the queue scheduling method of Based on Probability that the formation of node 3 and 5 correspondences is dispatched.Suppose that the formation for the treatment of output grouping of selecting is 5, then with formation 5 temporary transient formations as present node 13 correspondences.Then present node is moved to node 14, searching the child node of present node 14 in the fourth stage is 16 and 17, and all be not physical node, then adopt the strict-priority scheduling method that the formation of node 16 and 17 correspondences is dispatched, suppose that the priority of the formation of node 17 correspondences is higher than the priority of the formation of node 16 correspondences, the formation that is scheduled of therefore selecting this moment is the formation of node 17 correspondences.And node 17 is temporarily replaced by node 9 from the above, and then the formation that is scheduled selected this moment is actual is the formation of node 9 correspondences.Then present node is moved to node 10, this node does not have child node in the fourth stage, then skip this node, and present node is moved to node 15.The child node of node 15 in the fourth stage is 8,6, and is physical node, then adopts the queue scheduling method of Based on Probability that the formation of node 8 and 6 correspondences is dispatched.Suppose that the formation for the treatment of output grouping of selecting is 8, then with the formation 8 temporary transient present nodes 15 that replace.Finish the scheduling of the fourth stage this moment, when prime subtracts 1, namely three grades processed, and present node is moved to first node of the second level, and namely node 11.
Present node 11 is that node 13,14, two nodes are the sky node in the child node in prime, carries out strict priority scheduling.The priority of the formation of priority ratio 13 correspondences of the formation of node 14 correspondences is high, therefore formation corresponding to 14 nodes will be scheduled, and 14 nodes are temporarily replaced by No. 17 nodes, 17 nodes are replaced by 9 nodes again simultaneously, so the formation that in fact is scheduled is No. 9 formation corresponding to node, with No. 14 nodes of No. 9 temporary transient replacements of node.Present node is moved to node 12, and node 12 is node 10,15 in the child node in prime, and these two nodes are not physical node, therefore carries out strict priority scheduling.The priority of the formation that node 10 nodes are corresponding is higher than formation corresponding to 15 nodes, and therefore formation corresponding to 10 nodes will be scheduled, with No. 12 nodes of No. 10 temporary transient replacements of node.Finish the scheduling of the third level when this moment, when prime subtracts 1, namely secondary is processed, and present node is moved to first node of the first order, namely root node 0.
Present node 0 is that node 11,12, two all be empty node, so carry out strict priority scheduling in the child node in the prime.The priority of the formation of node 11 correspondences is higher than the priority of the formation of node 12 correspondences, and the formation that therefore is scheduled is the formation of node 11 correspondences.And node 11 is temporarily replaced by No. 9 nodes, and the formation that therefore is scheduled is No. 9 nodes, temporarily replaces root node 0 with No. 9 nodes.Because root node 0 is last node in the first order, therefore when the prime finishing scheduling.After subtracting 1, prime is the first order, so the epicycle finishing scheduling, the formation of node 9 correspondences is the formation that finally is scheduled.
It is pointed out that said process is to take turns scheduling process, obtain a formation that finally is scheduled of epicycle scheduling process after the finishing scheduling, can repeat the scheduling process that said process carries out next round, to obtain the next formation that finally is scheduled.
The present embodiment provides a kind of method of scheduling graded queues in packet network, by the performance index requirements value according to individual queue in the network formation is divided step by step, and priority query's group of formation tree structure, classification logotype according to the grouping that receives is included into affiliated formation afterbody with grouping, then according to priority query's group of tree structure, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in current priority query's group is dispatched; The present embodiment has solved the defective that can't take into account the network performance of higher priority queues and lower priority formation in the prior art, has realized having improved the network performance of Low Priority Queuing in the network performance of strict guarantee high-priority queue.
One of ordinary skill in the art will appreciate that: all or part of step that realizes said method embodiment can be finished by the relevant hardware of program command, aforesaid program can be stored in the computer read/write memory medium, this program is carried out the step that comprises said method embodiment when carrying out; And aforesaid storage medium comprises: the various media that can be program code stored such as ROM, RAM, magnetic disc or CD.
Fig. 4 is the structural representation of the device embodiment one of scheduling graded queues in packet network of the present invention, as shown in Figure 4, the present embodiment provides a kind of device of scheduling graded queues in packet network, specifically can carry out each step among the said method embodiment one, the device of the scheduling graded queues in packet network that the present embodiment provides can specifically comprise formation division module 401, grouping and classifying module 402 and queue scheduling module 403.Wherein, formation is divided module 401 and is used for according to the performance index requirements of individual queue described formation being divided step by step, and forms priority query's group of tree structure.Grouping and classifying module 402 is used for extracting the classification logotype of described grouping from the packets headers of the grouping that receives, the formation afterbody under according to described classification logotype described grouping being included into.Priority query's group that queue scheduling module 403 is used for according to described tree structure selects the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched.
Fig. 5 is the structural representation of the device embodiment two of scheduling graded queues in packet network of the present invention, as shown in Figure 5, the present embodiment provides a kind of device of scheduling graded queues in packet network, specifically can carry out each step among the said method embodiment two, the device of the scheduling graded queues in packet network that the present embodiment provides is on the basis of above-described embodiment one, can also specifically comprise statistical module 501, statistical module 501 is used for packet class and every kind of performance index requirements that grouping is corresponding of statistical packet network, and carries out buffer memory for every kind of grouping arranges independently equal-sized formation in buffering area.
Particularly, queue scheduling module 403 can specifically comprise and searches unit 413, the first scheduling unit 423, the second scheduling unit 433 and output acquiring unit 443.Wherein, search unit 413 for the child node of every one-level of searching tree structure and the subordinate relation of even higher level of node.The first scheduling unit 423 be used for when the child node number that be subordinated to same even higher level of node greater than 1, and described child node is physical node, then adopts the queue scheduling method of Based on Probability that the formation in current priority query's group is dispatched.The second scheduling unit 433 be used for when the son node number that be subordinated to same even higher level of node according to greater than 1, and described child node is not physical node, then adopts the Priority Queuing method that the formation in current priority query's group is dispatched; Wherein, described physical node is leaf node.Child node corresponding to the formation for the treatment of output grouping that output acquiring unit 443 is used for obtaining replaces the father node of described child node, every one-level from top to bottom to described tree structure repeats above-mentioned steps successively, replaced until the root node of described tree structure finally treats child node corresponding to the formation of output grouping, and from the formation of the child node representative that replaces described root node output grouping.
Further, the first scheduling unit 423 can comprise specifically that probability obtains subelement 4231 and subelement 4232 is obtained in formation.Wherein, probability obtain subelement 4231 be used for when the child node number that be subordinated to same even higher level of node greater than 1, and described child node according to the block length in the formation, and adopts formula (1) to set the scheduling probability of described formation when being physical node.Formation is obtained subelement 4232 and is used for obtaining the formation that described current priority query's group is treated output grouping according to scheduling probability and the random random number that generates of described formation.
Further, grouping and classifying module 402 can specifically comprise formation acquiring unit 412 and detecting unit 422.Wherein, formation acquiring unit 412 is used for obtaining formation under the described grouping according to described classification logotype.Detecting unit 422 if so, then abandons described grouping, otherwise described formation afterbody is included in described grouping for detection of the length the higher limit whether length of described formation equals to preset.
The present embodiment provides a kind of device of scheduling graded queues in packet network, by according to the performance index requirements of individual queue in the network formation being divided step by step, and priority query's group of formation tree structure, classification logotype according to the grouping that receives is included into affiliated formation afterbody with grouping, then according to priority query's group of tree structure, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in current priority query's group is dispatched; The present embodiment has solved the defective that can't take into account the network performance of higher priority queues and lower priority formation in the prior art, has realized having improved the network performance of Low Priority Queuing in the network performance of strict guarantee high-priority queue.
It should be noted that at last: above embodiment only in order to technical scheme of the present invention to be described, is not intended to limit; Although with reference to previous embodiment the present invention is had been described in detail, those of ordinary skill in the art is to be understood that: it still can be made amendment to the technical scheme that aforementioned each embodiment puts down in writing, and perhaps part technical characterictic wherein is equal to replacement; And these modifications or replacement do not make the essence of appropriate technical solution break away from the spirit and scope of various embodiments of the present invention technical scheme.
Claims (8)
1. the method for a scheduling graded queues in packet network is characterized in that, comprising:
Performance index requirements according to individual queue is divided step by step to described formation, and forms priority query's group of tree structure;
From the packets headers of the grouping that receives, extract the classification logotype of described grouping, the formation afterbody under according to described classification logotype described grouping being included into;
According to priority query's group of described tree structure, select the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched;
Described performance index requirements according to individual queue is divided step by step to described formation, and forms priority query's group of tree structure, specifically comprises:
First according to the scope of determining the network performance value in all formations from minimum network performance value to the highest network performance value, and this network performance value scope is divided into a series of little range sections, the network performance value is close and formation that belong to same range section is classified as same priority query group; The rest may be inferred, step by step the prioritization set of queues; The priority query's group that to the last once is divided into satisfies actual requirement;
Then the priority query's group that represents the maximum at all formation places with a root node, root node is father node, the several priority queries group that is divided into for the first time is expressed as corresponding child node, the child node of dividing for the first time is father node, and the priority query's group that is divided into for the second time is expressed as the child node of corresponding father node; The like, until all formations all are illustrated in the tree structure with the form of the leaf node of reality;
Described priority query's group according to described tree structure selects the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched, and specifically comprises:
Steps A: search the child node of every one-level of tree structure and the subordinate relation of even higher level of node, if be subordinated to the child node number of same even higher level of node greater than 1, and described child node is physical node, then adopts the queue scheduling method of Based on Probability that the formation in current priority query's group is dispatched; If be subordinated to the child node number of same even higher level of node greater than 1, and described child node is not physical node, then adopts the Priority Queuing method that the formation in current priority query's group is dispatched; Wherein, described physical node is leaf node;
Step B: child node corresponding to the formation for the treatment of output grouping that will obtain replaces the father node of described child node;
Every one-level from top to bottom to described tree structure repeats above-mentioned steps A and step B successively, replaced until the root node of described tree structure finally treats child node corresponding to the formation of output grouping, and from the formation of the child node representative that replaces described root node output grouping.
2. method according to claim 1 is characterized in that, also comprises:
Packet class in the statistical packet network and every kind of performance index requirements that grouping is corresponding, and in buffering area, carry out buffer memory for every kind of grouping arranges independently equal-sized formation.
3. method according to claim 1 is characterized in that, the queue scheduling method of described employing Based on Probability is dispatched the formation in current priority query's group and comprised:
According to the block length in the formation, and adopt following relational expression to set the scheduling probability of described formation:
Lm: Ln=Pn: Pm, wherein, Lm and Ln are respectively the block length among any two formation Qm and the Qn, and Pm and Pn are respectively the scheduling probability of described formation Qm and Qn;
Obtain the formation for the treatment of output grouping in described current priority query's group according to the scheduling probability of described formation and the random random number that generates.
4. method according to claim 1 is characterized in that, describedly according to described classification logotype the formation afterbody of described grouping under being included into is comprised:
Obtain formation under the described grouping according to described classification logotype;
Detect the length the higher limit whether length of described formation equals to preset, if so, then abandon described grouping, otherwise described formation afterbody is included in described grouping.
5. the device of a scheduling graded queues in packet network is characterized in that, comprising:
Module is divided in formation, is used for according to the performance index requirements of individual queue described formation being divided step by step, and forms priority query's group of tree structure;
Grouping and classifying module is used for extracting the classification logotype of described grouping, the formation afterbody under according to described classification logotype described grouping being included into from the packets headers of the grouping that receives;
The queue scheduling module is used for the priority query's group according to described tree structure, selects the queue scheduling method of strict-priority scheduling method or Based on Probability that the formation in the current described priority query group is dispatched;
Module is divided in described formation, the concrete scope that is used for determining to the highest network performance value from minimum network performance value according to all formations first the network performance value, and this network performance value scope is divided into a series of little range sections, the network performance value is close and formation that belong to same range section is classified as same priority query group; The rest may be inferred, step by step the prioritization set of queues; The priority query's group that to the last once is divided into satisfies actual requirement; Then the priority query's group that represents the maximum at all formation places with a root node, root node is father node, the several priority queries group that is divided into for the first time is expressed as corresponding child node, the child node of dividing for the first time is father node, and the priority query's group that is divided into for the second time is expressed as the child node of corresponding father node; The like, until all formations all are illustrated in the tree structure with the form of the leaf node of reality;
Described queue scheduling module specifically comprises searches unit, the first scheduling unit, the second scheduling unit and output acquiring unit:
Wherein, the described unit of searching is for the child node of every one-level of searching tree structure and the subordinate relation of even higher level of node;
Described the first scheduling unit, be used for when the child node number that be subordinated to same even higher level of node greater than 1, and described child node is physical node, then adopts the queue scheduling method of Based on Probability that the formation in current priority query's group is dispatched;
Described the second scheduling unit, be used for when the son node number that be subordinated to same even higher level of node according to greater than 1, and described child node is not physical node, then adopts the Priority Queuing method that the formation in current priority query's group is dispatched; Wherein, described physical node is leaf node;
Described output acquiring unit, child node corresponding to the formation for the treatment of output grouping that is used for obtaining replaces the father node of described child node;
Every one-level from top to bottom to described tree structure repeats described function of searching unit, the first scheduling unit, the second scheduling unit and output acquiring unit successively, until child node corresponding to the formation of finally treating output grouping that the root node of described tree structure is obtained by described output acquiring unit replace, and from the formation of the child node representative that replaces described root node output grouping.
6. device according to claim 5 is characterized in that, also comprises:
Statistical module is used for packet class and every kind of performance index requirements that grouping is corresponding of statistical packet network, and carries out buffer memory for every kind of grouping arranges independently equal-sized formation in buffering area.
7. device according to claim 5 is characterized in that, described the first scheduling unit comprises:
Probability obtains subelement, be used for when the child node number that be subordinated to same even higher level of node greater than 1, and described child node according to the block length in the formation, and adopts following relational expression to set the scheduling probability of described formation when being physical node:
Lm:Ln=Pn:Pm, wherein, Lm and Ln are respectively the block length among any two formation Qm and the Qn, and Pm and Pn are respectively the scheduling probability of described formation Qm and Qn;
Subelement is obtained in formation, is used for obtaining the formation that described current priority query's group is treated output grouping according to scheduling probability and the random random number that generates of described formation.
8. device according to claim 5 is characterized in that, described grouping and classifying module comprises:
The formation acquiring unit is used for obtaining formation under the described grouping according to described classification logotype;
Detecting unit, the length higher limit for detection of whether the length of described formation equals to preset if so, then abandons described grouping, otherwise described formation afterbody is included in described grouping.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010101585993A CN101848149B (en) | 2010-04-22 | 2010-04-22 | Method and device for scheduling graded queues in packet network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010101585993A CN101848149B (en) | 2010-04-22 | 2010-04-22 | Method and device for scheduling graded queues in packet network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101848149A CN101848149A (en) | 2010-09-29 |
| CN101848149B true CN101848149B (en) | 2013-10-23 |
Family
ID=42772613
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010101585993A Expired - Fee Related CN101848149B (en) | 2010-04-22 | 2010-04-22 | Method and device for scheduling graded queues in packet network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101848149B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2725750B1 (en) | 2011-08-04 | 2017-11-08 | Huawei Technologies Co., Ltd. | Queue scheduling method and device |
| CN103514037B (en) * | 2012-06-21 | 2018-06-01 | 中兴通讯股份有限公司 | Task scheduling processing method and device |
| CN102868633B (en) * | 2012-09-12 | 2015-04-29 | 中国联合网络通信集团有限公司 | Grouped data scheduling method and system |
| CN110618859B (en) * | 2019-09-10 | 2021-11-16 | 杭州秘猿科技有限公司 | Transaction pool design method supporting sequential packaging and electronic equipment |
| CN112787956B (en) * | 2021-01-30 | 2022-07-08 | 西安电子科技大学 | Method, system, storage medium and application for crowding occupation processing in queue management |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0782300A1 (en) * | 1995-12-27 | 1997-07-02 | Digital Equipment Corporation | Method and apparatus for hierarchical relative error scheduling |
| US20070086470A1 (en) * | 2005-10-03 | 2007-04-19 | Fujitsu Limited | Queue selection method and scheduling device |
| CN101272349A (en) * | 2008-05-06 | 2008-09-24 | 北京星网锐捷网络技术有限公司 | Priority level analog queue control method and device of quality service |
| CN101415242A (en) * | 2007-10-16 | 2009-04-22 | 中兴通讯股份有限公司 | System and method for scheduling wideband wireless access non-real time business |
| CN101557340A (en) * | 2009-05-07 | 2009-10-14 | 中兴通讯股份有限公司 | Method for realizing multilevel queue scheduling in data network and device |
| CN101631074A (en) * | 2009-07-30 | 2010-01-20 | 北京星网锐捷网络技术有限公司 | Multilink message sending method, device and network equipment |
-
2010
- 2010-04-22 CN CN2010101585993A patent/CN101848149B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0782300A1 (en) * | 1995-12-27 | 1997-07-02 | Digital Equipment Corporation | Method and apparatus for hierarchical relative error scheduling |
| US20070086470A1 (en) * | 2005-10-03 | 2007-04-19 | Fujitsu Limited | Queue selection method and scheduling device |
| CN101415242A (en) * | 2007-10-16 | 2009-04-22 | 中兴通讯股份有限公司 | System and method for scheduling wideband wireless access non-real time business |
| CN101272349A (en) * | 2008-05-06 | 2008-09-24 | 北京星网锐捷网络技术有限公司 | Priority level analog queue control method and device of quality service |
| CN101557340A (en) * | 2009-05-07 | 2009-10-14 | 中兴通讯股份有限公司 | Method for realizing multilevel queue scheduling in data network and device |
| CN101631074A (en) * | 2009-07-30 | 2010-01-20 | 北京星网锐捷网络技术有限公司 | Multilink message sending method, device and network equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101848149A (en) | 2010-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101848149B (en) | Method and device for scheduling graded queues in packet network | |
| CN101446962B (en) | Data conversion method, device thereof and data processing system | |
| KR101110904B1 (en) | Methods and apparatus for performing channel tree operations | |
| Hordijk et al. | On the assignment of customers to parallel queues | |
| US7472095B2 (en) | Methods and apparatus for automatic classification of text messages into plural categories | |
| US7979155B2 (en) | Sort plan optimization | |
| CN109033462A (en) | The method and system of low-frequency data item are determined in the storage equipment of big data storage | |
| CN105099729A (en) | User ID (Identification) recognition method and device | |
| CN105471670A (en) | Flow data classification method and device | |
| CN106909575A (en) | Text clustering method and device | |
| CN104125146A (en) | Service processing method and service processing method | |
| CN100488173C (en) | A method for carrying out automatic selection of packet classification algorithm | |
| CN103037375A (en) | Method and device for dividing community telephone traffic scenes | |
| CN101799803B (en) | Method, module and system for processing information | |
| CN100473229C (en) | Method for ascertaining removed channel in CDMA colony communication system | |
| CN111680046B (en) | Fast storage method for high-frequency power data | |
| CN104462448A (en) | Group name classification method and device | |
| CN115033646A (en) | Method for constructing real-time warehouse system based on Flink and Doris | |
| CN111090624B (en) | MES and CR plate-type customized furniture plate classification method | |
| CN101827175A (en) | Method and system for storing sorted call bills by catalog | |
| CN109447490B (en) | User address-based abnormal change relation discrimination method | |
| CN114416785A (en) | Stream type enterprise big data processing method and storage medium | |
| CN109240988B (en) | Method and system for preventing big data storage system from entering access imbalance state | |
| Nikolopoulos et al. | An experimental analysis of event set algorithms for discrete event simulation | |
| CN117372210B (en) | Legal service consultation 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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20131023 Termination date: 20140422 |