WO2014000569A1

WO2014000569A1 - Queue scheduling method and system

Info

Publication number: WO2014000569A1
Application number: PCT/CN2013/077072
Authority: WO
Inventors: 袁蕴超
Original assignee: 华为技术有限公司
Priority date: 2012-06-29
Filing date: 2013-06-09
Publication date: 2014-01-03
Also published as: CN102752202B; CN102752202A

Abstract

Disclosed are a queue scheduling method and system. The method comprises: selecting a position of a current scheduling pointer, and outputting a queue in a sub-scheduling unit at the position of the current scheduling pointer once; determining whether a start pointer camps; when it is determined that the start pointer camps, moving the start pointer to the position of the current scheduling pointer, and when it is determined that the start pointer does not camp, moving the start pointer to a position of a sub-scheduling unit following the position of the current scheduling pointer along a direction from the position of the start pointer to a position of a tail pointer; and after moving the start pointer, picking a position of a next scheduling pointer, and outputting a queue in a sub-scheduling unit at the position of the next scheduling pointer once. The method is capable of implementing flexible scheduling based on the queue situation of sub-scheduling units to be scheduled, so as to solve the problems of flow jitter and scheduling unevenness during RR pooling scheduling as well as the deterioration of the quality of service resulting from the scheduling unevenness in the prior art.

Description

The present invention relates to the field of communications technologies, and in particular, to a queue scheduling method and system.

BACKGROUND OF THE INVENTION QOS (quality of service) refers to a network communication process in which a network device needs to ensure a user's expected service level in terms of packet loss rate, delay, jitter, and bandwidth. This feature plays an important role in the allocation of network resources by operators, refining service levels, and improving operating profit. The TM (Traffic Management) chip in the network acts as a chip for managing user bandwidth, and is responsible for scheduling all user queues, and its scheduling algorithm has an important impact on the performance of the QOS service.

The existing scheduling algorithm maintains two tables, namely: an active table and an appending table, and the scheduling process is as shown in FIG. 1. For the scheduling unit in the active table, each scheduling unit corresponds to a token (token) bucket, The queue is not empty, the token bucket is not empty, and the deficit (deficit) > 0 has the scheduled permissions. The RR (Round robin) queues the scheduling unit and dequeues in sequence. During scheduling, when the scheduling unit dequeues once, the corresponding token bucket and deficit are subtracted from the dequeue data length. If defiitit<0 is detected, the scheduling unit moves to the appending table, and the decifix of the scheduling unit increases the weight x Quantum. Here weight is the scheduling weight of the current scheduling unit, and quantum is a fixed value, generally greater than 10000. If there is no schedulable unit in the current active table (that is, all scheduling units are moved to the appending table), or because token<0 cannot be dequeued, the active/appending switching operation is performed. The current appending table becomes the active table, and the current active table becomes the appending table.

Existing TM chips can support more than a thousand queues, even more than 256k, and also need to support hierarchical QOS, multiple levels of scheduling queues. Under this demand, the scheduler and queue management templates must be separated, so there is a time lag between scheduling and final data dequeue. In some specific scheduling situations, for example, each time the scheduling is a relatively long message, then after learning the length of the dequeued message, it is found that the deficit or the token is less than 0. Even if the scheduling of the current scheduling unit is stopped immediately, and the generated schedule continues to take effect, a deficit or token overflow occurs.

In order to prevent the deficit or token overflow, and to prevent the burst from scheduling longer data, the existing scheduling algorithm usually has a backpressure on the scheduler, that is, there must be a bandwidth-limited token bucket at the total exit, so that When the flow of the team is too large, or when the data is dispatched for a long period of time, the back pressure scheduler stops scheduling, and at the same time, the schedule that is less than the back pressure during the back pressure valid period is invalidated. Through the research on the prior art, the inventor found that: Since the scheduling mode of the RR polling is sequential scheduling, in a round of RR polling, each scheduling unit can only get a scheduling opportunity, so that the existing one There are several problems with the queue algorithm:

1), when the packet length of some scheduling units is relatively long, and the lengths of other scheduling units are relatively short, then in scheduling, the scheduling unit of the long message will obtain a larger dequeue bandwidth until it is selfed. The deficit uses the light, and then it can only schedule the shorter scheduling unit. In this case, if the preset traffic is relatively small and the weight is relatively large, the active/appending switching time of one round will be longer, resulting in a larger traffic jitter.

2) When all the scheduling units are long packets, the scheduling of one scheduling unit takes effect at this time. When the team is successfully dequeued, it will inevitably cause back pressure scheduling, and the following scheduling unit is dispatched. It was cancelled, which wasted an opportunity for RR polling. Although it is possible to solve the problem of scheduling backpressure through the method of buffer scheduling, usually the scheduling result of the large-scale user queue needs to return the scheduling result at a fixed time to refresh the queue state, so the scheduling cannot be generally cached. In the case of a large number of queues, a round of active/appending switching will take a long time, and if the deficit is sufficient, it is possible to always cancel the scheduling of some scheduling units, resulting in uneven scheduling. At the same time, because some users of large-scale user queues can't get scheduled, it will cause difficulty in queue management, causing the user's connection to be disconnected and the service quality to deteriorate.

SUMMARY OF THE INVENTION In view of this, the embodiments of the present invention provide a queue scheduling method and system, which can perform flexible scheduling according to a queue condition of a sub-scheduling unit to be scheduled, so as to solve the existing traffic jitter when using RR polling scheduling. Scheduling unevenness and the problem of poor quality of service due to uneven scheduling.

In order to achieve the above objectives, the technical solutions provided by the embodiments of the present application are as follows:

A queue scheduling method, in which all sub-scheduling units in the to-be-scheduled range are sequentially numbered, and the initial position of the starting pointer and the initial position of the tail pointer are selected within the to-be-tuned range, including:

Selecting the position of the first valid sub-scheduling unit between the initial position of the current starting pointer and the initial position of the tail pointer and pointing to the position of the tail pointer along the starting pointer position as the position of the scheduling pointer, and the current scheduling The queue in the sub-scheduling unit where the pointer is located is output once;

Determining, according to the queue length of the sub-scheduling unit where the current scheduling pointer is located, whether the starting pointer is resident, and the residing means that the starting pointer moves to a location where the current scheduling pointer is located;

When it is determined that the start pointer resides, move the position of the start pointer to the position of the current dispatch pointer, and select between the start pointer position and the tail pointer position after the shift, and after the shift The starting pointer position points to the position of the first valid sub-scheduling unit in the direction of the tail pointer position as the next scheduling pointer, and will be next The queue in the sub-scheduling unit where the dispatch pointer is located is output once;

When it is determined that the start pointer does not reside, moving the position of the start pointer to a position of a sub-scheduling unit after the position of the current dispatch pointer in the direction from the start pointer position to the tail pointer position, and selecting After shifting, between the start pointer position and the tail pointer position, and after the shift, the start pointer position points to the position of the first valid sub-scheduling unit in the direction of the tail pointer position as the position of the next dispatch pointer, and The queue in the sub-scheduling unit at the location where the next dispatch pointer is located is output once.

A scheduling system, in which all sub-scheduling units within the to-be-scheduled range are sequentially numbered, and the initial position of the starting pointer and the initial position of the tail pointer have been selected in the to-be-scheduled range, including:

a scheduling pointer position selecting unit, configured to select a position of a first valid sub-scheduling unit between the initial position of the current starting pointer and the initial position of the tail pointer, and the direction of the starting pointer position to the tail pointer position as the current scheduling pointer Location

a resident determining unit, configured to determine, according to a state of a sub-scheduling unit at a location of the current scheduling pointer, whether the starting pointer resides, where the residing refers to moving the starting pointer to the current scheduling pointer a start pointer shifting unit, configured to: when determining that the start pointer resides, move the position of the start pointer to a position where the current dispatch pointer is located, or when determining that the start pointer does not reside And moving the position of the start pointer to a position of one sub-scheduling unit after the position of the current dispatch pointer in the direction from the start pointer position to the tail pointer position;

And an output unit, configured to output the queue in the sub-scheduling unit where the current scheduling pointer is located. It can be seen from the foregoing technical solution that the queue scheduling method provided by the embodiment of the present application determines the state of the sub-scheduling unit at the location of the current scheduling pointer after scheduling the sub-scheduling unit at the location where the current scheduling pointer is located. Whether the pointer resides or not, when it is judged that the starting pointer resides, the child scheduling unit at the position where the current scheduling pointer is located is scheduled again. That is, the method can determine the number of scheduling according to the queue length of the sub-scheduling unit at the current scheduling pointer position, so that when scheduling, it is no longer scheduled for each sub-scheduling unit, and one of the scheduling directions can be treated. The sub-scheduling unit is scheduled multiple times.

Therefore, the embodiment of the present application provides the queue scheduling method, which can perform flexible scheduling according to the queue condition of the sub-scheduling unit to be scheduled, so as to solve the existing traffic jitter, uneven scheduling, and uneven scheduling caused by the RR polling scheduling. The problem that leads to poor service quality.

BRIEF DESCRIPTION OF THE DRAWINGS In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the embodiments or the prior art description will be briefly described below, and obviously, in the following description The drawing is only this Some of the embodiments described in the application can be obtained by those of ordinary skill in the art from the drawings without departing from the drawings.

FIG. 1 is a schematic flowchart of a queue scheduling method according to an embodiment of the present disclosure;

FIG. 2 is a schematic flowchart of a detailed process of determining whether a start pointer resides according to an embodiment of the present disclosure; FIG. 3 is a schematic flowchart of determining a value of a resident parameter according to a weight value according to an embodiment of the present disclosure; A detailed flow diagram of resetting the value of the resident parameter according to the scheduled back pressure provided by the example;

FIG. 5 is a schematic flowchart of another queue scheduling method according to an embodiment of the present disclosure;

FIG. 6 is a schematic structural diagram of a queue scheduling system according to an embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of a resident determining unit according to an embodiment of the present disclosure;

FIG. 8 is a schematic structural diagram of a resident computing unit according to an embodiment of the present disclosure;

FIG. 9 is a schematic structural diagram of a resident parameter adjustment unit according to an embodiment of the present application;

FIG. 10 is a schematic structural diagram of another queue scheduling system according to an embodiment of the present application.

DETAILED DESCRIPTION OF THE EMBODIMENTS In order to make those skilled in the art better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present application. The described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are intended to be within the scope of the application. An embodiment:

FIG. 1 is a schematic flowchart of a queue scheduling method according to an embodiment of the present disclosure.

Before scheduling, all sub-scheduling units within the to-be-scheduled range are sequentially set with numbers, and the initial position of the starting pointer and the initial position of the tail pointer are pre-selected within the to-be-scheduled range.

As shown in Figure 1, the method includes the following steps:

S 100: Select the position of the scheduling pointer, and output the queue in the sub-scheduling unit where the current scheduling pointer is located.

When the scheduling pointer is selected, the position of the first valid sub-scheduling unit located between the current starting pointer position and the tail pointer position and pointing to the tail pointer position along the starting pointer position is selected as the position of the current scheduling pointer.

Here, between the current start pointer position and the tail pointer position, the current start pointer and the current tail pointer are included, that is, If the sub-scheduling unit of the current starting pointer position is valid, then when the scheduling pointer is selected for the first time, the current starting pointer position is selected as the position of the scheduling pointer, that is, the position of the scheduling pointer and the current starting pointer will coincide.

S200: Determine whether the start pointer resides according to the queue length of the sub-scheduling unit where the current scheduling pointer is located.

Residing here means that the start pointer moves to the position of the current dispatch pointer, and the start pointer is moved to the position of the current dispatch pointer, so that the position of the start pointer coincides with the position of the current dispatch pointer, then the next time the dispatch pointer is selected According to the above method for selecting a dispatch pointer, the position of the current dispatch pointer will still be selected as the position of the next scheduled pointer, that is, by determining whether the start pointer is resident, it can be determined whether the sub-scheduling of the current pointer position is required. The unit performs repeated scheduling.

When it is judged in this step that the start pointer is resident, S300 is performed; when it is judged in this step that the start pointer is not resident, S400 is performed.

S300: Move the position of the start pointer to the position of the current dispatch pointer.

It has been described in the above S200 that the purpose of the relocation is to make the position of the start pointer coincide with the position of the current dispatch pointer. In the embodiment of the present application, if the location of the current scheduling pointer does not coincide with the location of the current starting pointer, when determining that the starting pointer resides, the location of the current starting pointer needs to be moved to the current scheduling pointer. position.

In addition, if the position of the current dispatch pointer and the position of the current start pointer are originally coincident, when the start pointer is determined to be resident, the position of the current start pointer can be directly maintained without moving, Of course, it can also be considered that the moving position of the current starting pointer is 0.

After moving the position of the start pointer, proceed to S500.

S400: Move the position of the start pointer to the position of one of the sub-scheduling units after the position of the current dispatch pointer in the direction of the start pointer position to the tail pointer position.

If it is judged that the starting pointer does not reside, that is, the sub-scheduling unit at the position where the current scheduling pointer is located does not need to be repeatedly scheduled, then the next sub-scheduling needs to be scheduled along the initial position of the starting pointer to the initial position of the tail pointer. unit.

When shifting, the position of the start pointer is moved to the position of the current dispatch pointer and then shifted to the position of the sub-scheduling unit, so that the next time the position of the dispatch pointer is selected, the sub-scheduling unit after the current dispatch pointer is necessarily selected.

After moving the position of the start pointer, proceed to S500.

S500: Select the position of the next scheduling pointer, and output the queue in the sub-scheduling unit where the next scheduling pointer is located. When the position of the next scheduling pointer is selected, the selection principle is the same as that in the above S100, and the first position between the pointer position and the tail pointer position after the shift is selected, and the position of the starting pointer position after the shift is directed to the direction of the tail pointer position. A valid sub-scheduling unit is used as the location of the next dispatch pointer.

After the start pointer is shifted, the position of the start pointer will coincide with the position of the current dispatch pointer. Therefore, the position of the next scheduled pointer is still the current dispatch pointer, that is, the scheduling is repeated for a certain scheduling subunit.

The queue scheduling method provided by the embodiment of the present application determines whether the start pointer is resident or not after scheduling the child scheduling unit at the location of the current scheduling pointer. When it is judged that the start pointer resides, the child scheduling unit at the position where the current dispatch pointer is located is scheduled again. That is, the method can determine the number of scheduling according to the queue length of the sub-scheduling unit at the current scheduling pointer position, so that when scheduling, it is no longer scheduled for each sub-scheduling unit, and one of the scheduling directions can be treated. The sub-scheduling unit is scheduled multiple times.

Therefore, the embodiment of the present application provides the queue scheduling method, which can perform flexible scheduling according to the queue condition of the sub-scheduling unit to be scheduled, so as to solve the existing traffic jitter, uneven scheduling, and scheduling due to the RR polling scheduling. Uniformity causes problems in service quality. Yet another embodiment:

In the embodiment of the present application, in order to facilitate the description of the queue length of the sub-scheduling unit, the queue lengths of all the sub-scheduling units in the to-be-scheduled range are represented by four states, namely: a first state, a second state, and a third State and fourth state, where:

The first state indicates that the sub-scheduling unit is empty; the second state indicates that the sub-scheduling unit has a small number of packets, and there may be a situation in which the sub-scheduling unit may be emptied; and the third state indicates that the sub-scheduling unit may continue in the case of no scheduling backpressure Scheduling; fourth state, indicating that the sub-scheduling unit has overflowed.

In practical applications, except for the first state and the fourth state, the other two states can be adjusted according to actual conditions.

In addition, in order to facilitate the presence of a more quantified representation of whether the start pointer resides, in the embodiment of the present application, the resident parameter is used to indicate whether the start pointer resides, specifically:

When the resident parameter is the first value, it means that the starting pointer does not reside; when the resident parameter is the second value, it means that the starting pointer resides.

FIG. 2 is a schematic flowchart of determining whether a start pointer resides according to an embodiment of the present application.

As shown in Figure 2, determining whether the starting pointer resides includes the following steps: S201: Determine the status of the sub-scheduling unit where the current scheduling pointer is located.

The state of the sub-scheduling unit is determined according to the length of the queue in the sub-scheduling unit where the current scheduling pointer is located. S202: Set the resident parameter to the first value.

When the state of the sub-scheduling unit at the position where the current scheduling pointer is located is the first state or the second state, the resident parameter is set to the first value;

S203: Determine whether the child scheduling unit at the location where the current scheduling pointer is located is scheduled for the first time.

When the sub-scheduling unit at the location where the current scheduling pointer is located is in the third state, it is determined whether the sub-scheduling unit at the location of the current scheduling pointer is scheduled for the first time. Determining whether the result of the first scheduling unit of the current scheduling pointer is scheduled may be used as a basis for scheduling the sub-scheduling unit at the location of the current scheduling pointer.

And when it is not the first scheduling, S2031 is performed; when it is the first scheduling, S2032 is performed.

S2031: Set the value of the resident parameter to the first value.

S2032: Determine the value of the resident parameter based on the weight value.

In this embodiment of the present application, as shown in FIG. 3, the step may include:

S301: Determine a random number range including a weight value according to the weight value, and randomly generate a random number in the random number range;

S302: Comparing the value of the random number with the weight value;

S303: When the value of the random number is less than or equal to the weight value, set the value of the resident parameter to the second value;

S304: When the value of the random number is greater than the weight value, the value of the resident parameter is set to the first value. S204: Set the value of the resident parameter to the second value, and reset the value of the resident parameter according to the scheduled back pressure during the scheduling process.

When the sub-scheduling unit at the position where the current scheduling pointer is located is in the fourth state, the value of the resident parameter is set to the second value, and the value of the resident parameter is reset according to the scheduling back pressure during the scheduling process.

In the embodiment of the present application, as shown in FIG. 4, resetting the value of the resident parameter according to the scheduling backpressure during the scheduling process may include the following steps:

S401: Compare the number of times the sub-scheduling unit at the location of the current scheduling pointer is scheduled with a preset maximum value, and determine whether a scheduling backpressure occurs.

The preset maximum value is less than or equal to the ratio of the longest packet length scheduled in the current sub-scheduling unit to the shortest packet length scheduled from the current sub-scheduling unit.

S402: If the number of times that the sub-scheduling unit at the current scheduling pointer is scheduled is less than a preset maximum value, and no scheduling back pressure occurs, determining a value of the resident parameter according to the weight value. S403: Set the value of the resident parameter to the first value.

If the number of times the sub-scheduling unit at the current scheduling pointer is scheduled is greater than or equal to the preset maximum value or any one of the scheduled back pressures occurs, the value of the resident parameter is set to the first value.

In the embodiment of the present application, how to determine whether to repeat scheduling the sub-scheduling unit at the location where the current scheduling pointer is located is determined according to the queue length in the sub-scheduling unit where the current scheduling pointer is located. Therefore, the flexible scheduling can be performed according to the queue condition of the sub-scheduling unit to be scheduled, so as to solve the problem that the existing traffic jitter, scheduling unevenness, and service quality deteriorate due to uneven scheduling when using RR polling scheduling. Yet another embodiment:

On the basis of the above two embodiments, when the start pointer moves along the start pointer and the direction of the tail pointer, the start pointer and the tail pointer will eventually coincide. When the start pointer and the tail pointer coincide, the present application The embodiment provides another queue scheduling method, so that all sub-scheduling units within the scheduling range can be re-scheduled.

FIG. 5 is a schematic flowchart diagram of another queue scheduling method according to an embodiment of the present disclosure.

As shown in FIG. 5, the queue scheduling method may further include:

S600: Determine whether the position of the start pointer coincides with the position of the tail pointer.

As described above, when the start pointer moves along the start pointer to the tail pointer direction, the start pointer and the tail pointer will eventually coincide. Therefore, during the scheduling process, it is necessary to determine whether the start pointer is the same as the tail pointer. coincide. S700: Move the start pointer to the initial position of the start pointer when the position of the start pointer coincides with the position of the tail pointer.

Moving the start pointer to its initial position means that all sub-scheduling units within the scheduling range can be scheduled again.

The method provided by the embodiment of the present application may move the position of the start pointer to the initial position of the start pointer after one scheduling is completed, and then all the child scheduling units may be scheduled again after one scheduling is completed. Yet another embodiment:

FIG. 6 is a schematic structural diagram of a queue scheduling system according to an embodiment of the present application.

As shown in FIG. 6, the queue scheduling system includes: a scheduling pointer location selecting unit 1, a resident determining unit 2, The pointer shift unit 3 and the output unit 4 are started.

The dispatch pointer position selecting unit 1 is configured to select the position of the first valid sub-scheduling unit between the initial position of the current starting pointer and the initial position of the tail pointer, and the direction of the starting pointer position to the tail pointer position as the current scheduling pointer. position.

Here, between the current start pointer position and the tail pointer position, the current start pointer and the current tail pointer are included, that is, if the child scheduling unit of the current start pointer position is valid, then when the schedule pointer is selected for the first time, it is selected. The current starting pointer position is used as the position of the dispatch pointer, that is, the position of the dispatch pointer and the current start pointer will coincide.

The resident judging unit 2 is configured to determine, according to the state of the sub-scheduling unit at the location of the current scheduling pointer, whether the starting pointer resides after the scheduling pointer location selecting unit 1 selects the location of the current scheduling pointer.

By judging whether the start pointer is resident, it can be determined whether it is necessary to perform repeated scheduling on the sub-scheduling unit where the current pointer position is located.

The start pointer shifting unit 3 is configured to move the position of the start pointer to the position of the current dispatch pointer when the resident judgment unit determines that the start pointer resides.

The purpose of the relocation is to make the position of the starting pointer coincide with the position of the current dispatch pointer. In the embodiment of the present application, if the location of the current scheduling pointer does not coincide with the location of the current starting pointer, when determining that the starting pointer resides, the location of the current starting pointer needs to be moved to the current scheduling pointer. position.

The output unit 4 is configured to output the queue of the sub-scheduling unit at the location of the current scheduling pointer once after selecting the location of the current scheduling pointer.

After scheduling the sub-scheduling unit at the location of the current scheduling pointer, the system determines whether the starting pointer resides on the state of the sub-scheduling unit at the location of the current scheduling pointer, and when determining that the starting pointer resides, The sub-scheduling unit at the location where the current dispatch pointer is located is scheduled. That is, the method can determine the number of scheduling according to the queue length of the sub-scheduling unit at the current scheduling pointer position, so that when scheduling, it is no longer scheduled for each sub-scheduling unit, and one of the scheduling directions can be treated. The sub-scheduling unit is scheduled multiple times. Yet another embodiment:

In the embodiment of the present application, in order to facilitate the description of the queue length of the sub-scheduling unit, the queue lengths of all the sub-scheduling units in the to-be-scheduled range are represented by four states, namely: a first state, a second state, and a third State And a fourth state, where:

FIG. 7 is a schematic structural diagram of a resident determination unit according to an embodiment of the present application.

As shown in FIG. 7, in the embodiment of the present application, the resident determining unit 2 includes: a state determining unit 21, a first scheduling determining unit 22, a resident probability calculating unit 23, a resident parameter value setting unit 24, and a resident. Parameter adjustment unit 25.

The state determining unit 21 is for determining the state of the child scheduling unit at the location where the current scheduling pointer is located. When determining, the state of the sub-scheduling unit is determined according to the length of the queue in the sub-scheduling unit where the current scheduling pointer is located.

The first scheduling determining unit 22 is configured to determine whether to schedule the sub-scheduling unit at the location of the current scheduling pointer for the first time. Determining whether the result of the first scheduling of the sub-scheduling unit at the location where the current scheduling pointer is located may be used as a basis for scheduling the sub-scheduling unit at the location of the current scheduling pointer.

The resident probability calculation unit 23 is configured to determine, based on the weight value, whether the start pointer resides at the position of the current start pointer.

In the embodiment of the present application, as shown in FIG. 8, the resident calculation unit 23 may include: a random number generation unit 31, a second comparison unit 32, and a resident probability calculation sub-unit 33, where:

The random number range generating unit 31 is configured to determine a random number range including the weight value according to the weight value, and randomly generate a random number in the random number range;

The second comparison unit 32 is configured to compare the value of the random number with the weight value;

The resident probability calculation sub-unit 33 is configured to calculate whether the start pointer resides according to the comparison result of the second comparison unit 32, and when the value of the random number is less than or equal to the weight value, indicating that the start pointer resides; when the value of the random number is greater than the weight A value indicating that the starting pointer does not reside.

The resident parameter value setting unit 23 is configured to set the resident parameter to the first value when the state of the sub-scheduling unit at the position where the current scheduling pointer is located is the first state or the second state;

When the sub-scheduling unit at the location where the current scheduling pointer is located is in the third state, and is not the first scheduling, the value of the resident parameter is set to the first value;

When the sub-scheduling unit of the current scheduling pointer is in the third state, and is the first scheduling, the resident parameter is set to the first value or the second value according to the calculation result of the resident probability calculation unit; When the sub-scheduling unit at the position where the current scheduling pointer is located is in the fourth state, the value of the resident parameter is set to the second value.

The resident parameter adjustment unit 25 is configured to set the value of the resident parameter to the second value when the sub-scheduling unit at the location where the current scheduling pointer is located, and reset the resident according to the scheduling back pressure during the scheduling process. Leave the value of the parameter.

In the embodiment of the present application, as shown in FIG. 9, the resident parameter adjustment unit 24 may include: a first comparison unit

41 and a scheduled back pressure detecting unit 42, wherein:

The first comparing unit 41 is configured to compare the number of times the sub-scheduling unit of the current scheduling pointer is scheduled with the preset maximum value;

The scheduling back pressure detecting unit 42 is configured to detect whether a back pressure occurs during the scheduling process.

When the number of times the sub-scheduling unit at the current scheduling pointer is scheduled is less than the preset maximum value and the scheduling back pressure does not occur, the resident parameter value setting unit 24 sets the resident parameter according to the calculation result of the resident probability calculation unit 23 to The first value or the second value, otherwise the value of the resident parameter is set to the first value.

The preset maximum value is less than or equal to the ratio of the longest packet length scheduled in the current sub-scheduling unit to the shortest packet length scheduled from the current sub-scheduling unit. Yet another embodiment:

As shown in FIG. 10, the queue scheduling system may further include: a position determining unit 5 and an initializing unit 6, wherein: the position determining unit 5 is configured to determine whether the position of the starting pointer coincides with the position of the tail pointer. As described above, when the start pointer moves along the start pointer to the tail pointer direction, the start pointer and the tail pointer will eventually coincide. Therefore, during the scheduling process, it is necessary to determine whether the start pointer is the same as the tail pointer. coincide.

The initializing unit 6 is configured to move the starting pointer to the initial position of the starting pointer after the position of the starting pointer coincides with the position of the tail pointer. Moving the starting pointer to its initial position means that all child scheduling units within the scheduling range can be scheduled again.

The system provided by the embodiment of the present application can move the position of the start pointer to the initial position of the start pointer after one scheduling is completed, and then all the child scheduling units can be scheduled again after one scheduling is completed. For the convenience of description, the above devices are described separately by function into various units. Of course, the functions of the various units may be implemented in one or more software and/or hardware in the implementation of the present application.

It will be apparent to those skilled in the art from the above description of the embodiments that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product, which may be stored in a storage medium such as a ROM/RAM or a disk. , an optical disk, etc., includes instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to perform the methods described in various embodiments of the present application or portions of the embodiments.

The various embodiments in the specification are described in a progressive manner, and the same or similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment. The system embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, ie may be located One place, or it can be distributed to multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the embodiment. Those of ordinary skill in the art can understand and implement without any creative effort.

The above description is only a preferred embodiment of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments are obvious to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the application. Therefore, the present invention is not limited to the embodiments shown herein, but the broadest scope consistent with the principles and novel features disclosed herein.

Claims

Rights request

1. A queue scheduling method in which all sub-scheduling units within the range to be scheduled are sequentially set with numbers, and the initial position of the start pointer and the initial position of the tail pointer have been selected within the range to be scheduled, which is characterized by including: selection The position of the first valid sub-scheduling unit between the initial position of the current start pointer and the initial position of the tail pointer, and along the starting pointer position pointing to the direction of the tail pointer position, is used as the position of the scheduling pointer, and the current scheduling pointer is The queue in the sub-scheduling unit at the location is output once;

Determine whether the start pointer is parked based on the queue length of the sub-scheduling unit where the current scheduling pointer is located. The parking means that the start pointer moves to the location of the current scheduling pointer;

When it is judged that the starting pointer is resident, the position of the starting pointer is moved to the position of the current scheduling pointer, and the position between the starting pointer position and the tail pointer position after the shift is selected, and the position along the shifted pointer is selected. The starting pointer position points to the first valid sub-scheduling unit in the direction of the tail pointer position as the position of the next scheduling pointer, and the queue in the sub-scheduling unit at the position of the next scheduling pointer is output once;

When it is judged that the start pointer does not reside, the position of the start pointer is moved to a position one sub-scheduling unit behind the current scheduling pointer position in the direction from the starting pointer position to the tail pointer position, and selects The first valid sub-scheduling unit between the starting pointer position and the tail pointer position after the shift and the starting pointer position pointing to the tail pointer position direction after the shift is the position of the next scheduling pointer, and The queue in the sub-scheduling unit where the next scheduling pointer is located is output once.

2. The method according to claim 1, characterized in that the queue length of all sub-scheduling units within the to-be-scheduled range is represented by four states, wherein: the first state indicates that the sub-scheduling unit is empty; the second state The state indicates that the sub-scheduling unit has a small number of messages and may be empty; the third state indicates that the sub-scheduling unit can be scheduled continuously without scheduling back pressure; the fourth state indicates that the sub-scheduling unit has overflowed;

Whether the start pointer is resident is indicated by the resident parameter, and when the resident parameter is the first value, it means that the start pointer is not resident; when the resident parameter is the second value , indicating that the start pointer is resident; and judging whether the start pointer is resident based on the state of the sub-scheduling unit where the current scheduling pointer is located specifically includes:

Determine the status of the sub-scheduling unit where the current scheduling pointer is located;

When the state of the sub-scheduling unit at the position of the current scheduling pointer is the first state or the second state, the resident parameter is set to the first value;

When the sub-scheduling unit at the position of the current scheduling pointer is in the third state, determine whether the sub-scheduling unit at the position of the current scheduling pointer is scheduled for the first time, and when it is not the first time to schedule, the resident The value of the parameter is set to the first value. When it is scheduled for the first time, the value of the resident parameter is determined based on the weight value; When the sub-scheduling unit where the current scheduling pointer is located is in the fourth state, the value of the dwell parameter is set to the second value, and the value of the dwell parameter is reset according to the scheduling backpressure during the scheduling process. .

3. The method according to claim 2, characterized in that, when the sub-scheduling unit at the position of the current scheduling pointer is in the fourth state, the value of the resident parameter is reset according to the scheduling back pressure during the scheduling process, Specifically, it includes: comparing the number of times the sub-scheduling unit at the current scheduling pointer position is scheduled with the preset maximum value, and determining whether scheduling back pressure occurs;

If the number of times the sub-scheduling unit at the current scheduling pointer position is scheduled is less than the preset maximum value and no scheduling backpressure occurs, the value of the dwell parameter is determined based on the weight value, otherwise the value of the dwell parameter is set. is the first value.

4. The method according to claim 2, wherein the preset maximum value is less than or equal to the ratio of the longest message length scheduled in the current sub-scheduling unit to the shortest message length scheduled from the current sub-scheduling unit.

5. The method according to claim 2 or 3, characterized in that determining the value of the resident parameter according to the weight value specifically includes:

Determine a random number range including the weight value according to the weight value;

Randomly generate a random number within the range of the random number;

Compare the value of the random number with the weight value;

When the value of the random number is less than or equal to the weight value, the value of the resident parameter is set to the second value; when the value of the random number is greater than the weight value, the value of the resident parameter is set to the second value. Set to the first value.

6. The method according to any one of claims 1 to 5, further comprising:

Determine whether the position of the starting pointer coincides with the position of the tail pointer;

When the position of the starting pointer coincides with the position of the tail pointer, the starting pointer is moved to the initial position of the starting pointer.

7. A scheduling system in which all sub-scheduling units within the range to be scheduled are sequentially set with numbers, and the initial position of the start pointer and the initial position of the tail pointer have been selected within the range to be scheduled, which is characterized by including:

The scheduling pointer position selection unit is used to select the position of the first valid sub-scheduling unit between the current initial position of the start pointer and the initial position of the tail pointer and along the direction of the starting pointer position pointing to the tail pointer position as the current scheduling pointer. Location;

A residence judgment unit, configured to determine whether the start pointer resides according to the state of the sub-scheduling unit where the current scheduling pointer is located. The residence means that the start pointer moves to where the current scheduling pointer is located. Position; a starting pointer shifting unit, used to move the position of the starting pointer to the position of the current scheduling pointer when it is judged that the starting pointer is resident, or when it is judged that the starting pointer is not resident. When , move the position of the starting pointer to one position behind the current scheduling pointer in the direction from the starting pointer position to the tail pointer position. The location of the dispatch unit;

An output unit is used to output the queue in the sub-scheduling unit where the current scheduling pointer is located once.

8. The system according to claim 6, characterized in that the queue length of all sub-scheduling units within the range to be scheduled is represented by four states, wherein: the first state indicates that the sub-scheduling unit is empty; the second state The state indicates that the sub-scheduling unit has a small number of messages and may be empty; the third state indicates that the sub-scheduling unit can be scheduled continuously without scheduling back pressure; the fourth state indicates that the sub-scheduling unit has overflowed;

Whether the start pointer is resident is indicated by the resident parameter, and when the resident parameter is the first value, it means that the start pointer is not resident; when the resident parameter is the second value , indicating that the start pointer resides; the residence judgment unit includes:

The status determination unit is used to determine the status of the sub-scheduling unit where the current scheduling pointer is located; the first scheduling judgment unit is used to determine whether the sub-scheduling unit where the current scheduling pointer is located is scheduled for the first time;

A residence probability calculation unit, used to determine whether the starting pointer resides at the position of the current starting pointer according to the weight value;

A resident parameter value setting unit, configured to set the resident parameter to a first value when the state of the sub-scheduling unit where the current scheduling pointer is located is the first state or the second state; when the current scheduling pointer is located When the sub-scheduling unit at the position is in the third state, and it is not the first time of scheduling, the value of the resident parameter is set to the first value; when the sub-scheduling unit at the position of the current scheduling pointer is in the third state When , and it is the first time of scheduling, the residency parameter is set to the first value or the second value according to the calculation result of the residency probability calculation unit; when the sub-scheduling unit where the current scheduling pointer is located is the fourth When in the state, set the value of the resident parameter to the second value;

A residence parameter adjustment unit, configured to set the value of the residence parameter to a second value when the sub-scheduling unit at the position of the current scheduling pointer is in the fourth state, and to perform the operation according to the scheduling back pressure during the scheduling process. Reset the value of the resident parameter.

9. The system according to claim 8, characterized in that, the resident parameter adjustment unit includes: a first comparison unit, used to compare the number of times the sub-scheduling unit at the current scheduling pointer position is scheduled with the preset maximum value Compare;

Scheduling back pressure detection unit, used to detect whether back pressure occurs during the scheduling process;

When the number of times the sub-scheduling unit at the position of the current scheduling pointer is scheduled is less than the preset maximum value and no scheduling backpressure occurs, the residency parameter value setting unit sets the residency parameter according to the calculation result of the residency probability calculation unit. Set to the first value or the second value, otherwise the value of the resident parameter is set to the first value.

10. The system according to claim 9, characterized in that the residence probability calculation unit includes: A random number range selection unit, used to determine a random number range including the weight value according to the weight value;

A random number generation unit, used to randomly generate a random number within the range of the random number;

A second comparison unit, configured to compare the value of the random number with the weight value;

A residence probability calculation subunit is used to calculate whether the start pointer is resident based on the comparison result of the second comparison unit. When the value of the random number is less than or equal to the weight value, it indicates that the start pointer is resident. stay; When the value of the random number is greater than the weight value, it means that the start pointer does not stay.

11. The system according to any one of claims 7-10, characterized in that, the system further includes: a position judgment unit, used to judge whether the position of the start pointer coincides with the position of the tail pointer; initialization A unit configured to move the start pointer to the initial position of the start pointer after checking whether the position of the start pointer coincides with the position of the tail pointer.