CN102082765A - User and service based QoS (quality of service) system in Linux environment - Google Patents
User and service based QoS (quality of service) system in Linux environment Download PDFInfo
- Publication number
- CN102082765A CN102082765A CN2009101994282A CN200910199428A CN102082765A CN 102082765 A CN102082765 A CN 102082765A CN 2009101994282 A CN2009101994282 A CN 2009101994282A CN 200910199428 A CN200910199428 A CN 200910199428A CN 102082765 A CN102082765 A CN 102082765A
- Authority
- CN
- China
- Prior art keywords
- formation
- message
- llq
- user
- queue
- 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.)
- Pending
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to a user and service based QoS (quality of service) system in a Linux environment, which comprises a filter rule library, a filter, a classification rule library, a classifier and a queue maintaining module, wherein the filter rule library is connected with the filter; the classification rule library is connected with the classifier; the filter, the classifier and the queue maintaining module are successively connected; and the queue maintaining module adopts a structure of multistage queue plus LLQ (low latency queuing) and adopts a DWRR (display window right register) algorithm for dispatching and sending messages. Compared with the prior art, the system provided by the invention has the advantages of reducing sources required for system realization to the minimum, ensuring the system stability and the like.
Description
Technical field
The present invention relates to a kind of QoS system, especially relate under a kind of Linux environment based on user and professional QoS system.
Background technology
The application of present IP network carrying multiple business, as multimedia service, IPTV business, VoIP business, P2P business, wherein the P2P business is particularly outstanding.Therefore, essential based on the flow control of business.Meanwhile, the network user of a large-scale local area network (LAN) inside must be divided into many grades according to different identity or position, and at the user of these different brackets, the distribution of its network bandwidth is also with difference.This demand needs a kind of more satisfactory framework to realize.
Linux kernel begins to support QoS from Kernel 2.1.105 version, its working mechanism produces when congested when network, according to user's configuration, sets up different output queues at data output end at different data flow, carry out different scheduling and processing, thereby realize flow control.This will solve the network congestion problem that adopts fifo fifo mechanism to produce owing to early stage IP network to a certain extent.Grouping enter after the receiving system that adopts Linux handling process as shown in Figure 1.
In linux system, according to system sequence of operations performed in the flow control process is carried out in grouping, the flow-control module of Linux can be divided into three big elements, as shown in Figure 2.The flow control of linux system is the organic assembling to following three big objects:
1) formation rule (Queue Discipline): the process of transmitting of the packet in the control formation;
2) classification (Classes): the packet that arrives the network equipment according to its QoS attribute (as speed, priority etc.), is divided into different classifications with packet;
3) filter (Filters): the header packet information (as IP address, ToS value, protocol type, port etc.) according to packet is classified to packet.
The dispatching algorithm of QoS has a variety of.Different dispatching algorithms has different algorithm thought and algorithm complex according to the difference of controlled target.It is common that to have dispatching algorithm to have following several:
1) fifo queue FIFO: this is a kind of from realizing the simplest queue scheduling algorithm of angle, and the order that enters formation according to grouping goes out team.
2) PQ of priority query: be grouped in the formation that has different priorities by being arranged behind the grader to enter into.Scheduling thought is the at first schedules high priority formation just formation of scheduling time high priority of not divided into groups in this formation.Therefore be also referred to as strict-priority.
3) Weighted Fair Queuing WFQ: will divide into groups to classify according to stream, each stream is arranged a formation and is composed a priority value, priority value according to formation carries out allocated bandwidth, and the system that makes can carry out refined control to single Business Stream, but also increases algorithm complex simultaneously.
4) weights circulation wheel changes WRR: will divide into groups to enter into different formations according to the business of different brackets, be the queue assignment weights according to the height of service priority, and each formation of poll during scheduling guarantees that each formation all obtains certain service time.
Though WFQ can carry out based on the meticulous control of flowing the message in the network, its algorithm complex also increases many relatively, and the algorithm of WRR is carried out the queue structure that thought relatively meets native system, and algorithm complex is also relatively simply many.But the WRR algorithm can embody operational performance preferably when long messages such as processing, then be stymied by for non-isometric message, and the message length in the network almost each is unequal, this just needs to have adopted the evolution algorithm DWRR of WRR, can not provide the shortcoming of accurate control to the formation of elongated message to overcome the WRR algorithm.
Summary of the invention
Purpose of the present invention is exactly to provide under a kind of Linux environment based on user and professional QoS system for the defective that overcomes above-mentioned prior art existence.
Purpose of the present invention can be achieved through the following technical solutions:
Under a kind of Linux environment based on user and professional QoS system, it is characterized in that, this system comprises filtering rule storehouse, filter, classifying rules storehouse, grader, formation maintenance module, described filtering rule storehouse is connected with filter, described classifying rules storehouse is connected with grader, described filter, grader, formation maintenance module connect successively, and described formation maintenance module adopts multi-queue+LLQ structure, adopt the DWRR algorithm when described formation maintenance module scheduling sends message.
Described filtering rule storehouse comprises five-tuple, ToS value information.
Described classifying rules storehouse comprises classification number, queue number.
Described multi-queue+LLQ structure is to increase a LLQ formation in three grades of formations of multi-queue, wherein the priority of this LLQ formation is higher than other any formations in three grades of formations, by token bucket is maximum bandwidth reserved of LLQ queue assignment, if network does not take place when congested, the message very first time that enters LLQ is forwarded, and network takes place when congested, the message that enters LLQ then will be by speed limit, message for the overshoot flow will be dropped, when system always preferentially sends message among the LLQ do not have message in LLQ or when surpassing maximum bandwidth reserved for the LLQ configuration, just dispatch the message of other formations in three grades of formations of transmission.
Described DWRR algorithm concrete steps are as follows:
1) for each formation is provided with a difference counter DC, the value of DC is exactly the total amount of byte that scheduler can send at every turn;
2) according to the size of formation weights, for each formation is provided with a quantity variable, what variable was represented is the byte traffic volume of distributing to formation;
3) scheduler is visited a non-empty queue at every turn, if the message byte number of formation head is greater than DC, scheduler moves to next formation so, if the message byte number of formation head is less than or equal to DC, scheduler is just transmitted this message and the value of DC reduced the byte number that changes message so;
4) continue this process, till the byte number of this formation mop message is greater than DC;
5) remaining value is accumulated to and uses next time, if formation is empty, then DC is set to 0, and at this moment scheduler will be served next non-empty queue.
Described scheduler is arranged in the formation maintenance module.
Compared with prior art, the present invention has the following advantages:
1) the formation maintenance module has adopted the design of multi-queue+LLQ, has realized that LLQ has also guaranteed the delay requirement of low deferred class business simultaneously to user and professional purpose of carrying out dual control;
Adopted the DWRR algorithm when 2) sending scheduling, fair relatively carries out scheduling according to service priority to grouping, has also guaranteed minimum algorithm complex simultaneously, and it is minimum to make system realize that resource requirement drops to, and has guaranteed the stability of system.
Description of drawings
Fig. 1 is based on Linux flow control procedure in the prior art of the user and the QoS system of business under a kind of Linux environment of the present invention;
Fig. 2 is based on linux kernel flow control framework in the prior art of the user and the QoS system of business under a kind of Linux environment of the present invention;
Fig. 3 is based on the structural framing figure of user with the QoS system of business under a kind of Linux environment of the present invention;
Fig. 4 is based on the structure chart of user with the multi-queue+LLQ of the QoS system of business under a kind of Linux environment of the present invention;
Fig. 5 is based on the DWRR algorithm schematic diagram of user with the QoS system of business under a kind of Linux environment of the present invention.
Embodiment
The present invention is described in detail below in conjunction with the drawings and specific embodiments.
Embodiment
As shown in Figure 3, under a kind of Linux environment based on user and professional QoS system, it is characterized in that, this system comprises filtering rule storehouse d, filter a, classifying rules storehouse e, grader b, formation maintenance module c, described filtering rule storehouse d is connected with filter a, described classifying rules storehouse e is connected with grader b, described filter a, grader b, formation maintenance module c connect successively, described formation maintenance module c adopts multi-queue+LLQ structure, adopts the DWRR algorithm when described formation maintenance module c scheduling sends message.Described filtering rule storehouse d comprises five-tuple, ToS value information.Described classifying rules storehouse e comprises classification number, queue number.
As shown in Figure 4, described formation maintenance module adopts multi-queue+LLQ structure, suppose all network users are divided three classes, the user organizes 1, organizes 2, organizes 3, wherein the user organizes 1 for the user of highest level, below promptly claim the power user, organize 3, below be referred to as domestic consumer and the user organizes 2 the user that is superior to.Above cohort design is to have considered user and professional double factor.
Because the power user is special user's group in the network, their relative domestic consumer enjoys higher bandwidth and guarantees, therefore, in design they and domestic consumer is differentiated, and for they open up a network channel, guarantees their user bandwidth whereby separately.At domestic consumer, they are with residual bandwidth resources in the shared network.Safeguard a set of queues in each user's group, represent dissimilar service application respectively, this is the one-level formation.Outside the unique user group, safeguard a set of queues in domestic consumer's group in addition, represent different application equally, this is the secondary formation of domestic consumer, according to the grouping in the different weight scheduling one-level formations, realize user's control of domestic consumer according to the rank of different user groups in the domestic consumer.The power user then is that portion safeguards one group of one-level formation and secondary formation respectively within it, and the secondary formation is dispatched according to different weights according to the rank of its one-level formation business, realizes power user's self business control.Outside all users organize, safeguard three grades of formations, dispatch separately secondary formation respectively, realize that power user's the user's control and the business of domestic consumer control according to the rank of power user's assurance bandwidth and domestic consumer's different business.Above design has realized user and professional purpose of carrying out dual control.
The present invention is directed to a class application layer business of in the practical application network delay characteristic being had relatively high expectations,, provide the low preferably service that postpones, with quality of service and the user satisfaction that guarantees this class business as the rtp streaming amount in the network (IPTV, VoIP etc.).The third level at described multi-queue increases a LLQ formation, opens up a network channel separately for low deferred traffic, only dispatches the low deferred traffic of power user and domestic consumer.
LLQ (Low Latency Queueing, Low Latency Queueing) is a formation with higher priority, and in described multi-queue, its priority is higher than other any formations in three grades of formations.System is that LLQ distributes a maximum bandwidth reserved, when message is scheduled out group, if when having among the LLQ message then always preferentially to send message among the LLQ in LLQ, not have message or when surpassing maximum bandwidth reserved for the LLQ configuration, just dispatch the message of other formations in other three grades of formations of transmission.When message entered LLQ, congested if network does not take place, then the message of LLQ can be sent out away in the very first time; And network takes place when congested, and the message that enters LLQ then will be by speed limit, will be dropped for the message of overshoot flow.Do not take place to make under the congested situation message that belongs to LLQ can obtain idle bandwidth at interface like this and exist, and the message that can guarantee to belong to LLQ under the situation of interface congestion can not take the bandwidth of overshoot, protect the deserved bandwidth of other messages.The configure reserved bandwidth can realize by a token bucket is set before the formation of LLQ.In addition because as long as there is messaging system will send message among the LLQ among the LLQ, so the delay that the message among the LLQ is sent out is at most the time that interface sends a maximum length message, no matter be time-delay or delay jitter LLQ can be reduced to bottom line with it, this is for providing the favorable service quality assurance to delay time responsive application such as VoIP business.
Adopt the DWRR algorithm when as shown in Figure 5, described formation maintenance module scheduling sends message.This algorithm is the algorithm of algorithm complex and handling property compromise, has guaranteed the operational performance of system high-speed and forwarding performance efficiently.DWRR is a kind of derivative algorithm of WRR, and it has overcome the WRR algorithm can not provide accurate control to the formation of elongated message shortcoming.Specific algorithm is as follows:
1) for each formation is provided with a difference counter DC, the value of DC is exactly the total amount of byte that scheduler can send at every turn;
2) according to the size of formation weights, for each formation is provided with a quantity variable, what variable was represented is the byte traffic volume of distributing to formation;
3) scheduler is visited a non-empty queue at every turn, if the message byte number of formation head is greater than DC, scheduler moves to next formation so, if the message byte number of formation head is less than or equal to DC, scheduler is just transmitted this message and the value of DC reduced the byte number that changes message so;
4) continue this process, till the byte number of this formation mop message is greater than DC;
5) remaining value is accumulated to and uses next time, if formation is empty, then DC is set to 0, and at this moment scheduler will be served next non-empty queue.
Owing to be provided with difference counter, though limited the byte quantity that it can send, only can influence the performance of this formation, and can not have influence on other formations, realized professional isolation.
Claims (6)
- Under the Linux environment based on user and professional QoS system, it is characterized in that, this system comprises filtering rule storehouse, filter, classifying rules storehouse, grader, formation maintenance module, described filtering rule storehouse is connected with filter, described classifying rules storehouse is connected with grader, described filter, grader, formation maintenance module connect successively, described formation maintenance module adopts multi-queue+LLQ structure, adopts the DWRR algorithm when described formation maintenance module scheduling sends message.
- 2. based on user and professional QoS system, it is characterized in that described filtering rule storehouse comprises five-tuple, ToS value information under a kind of Linux environment according to claim 1.
- 3. based on user and professional QoS system, it is characterized in that described classifying rules storehouse comprises classification number, queue number under a kind of Linux environment according to claim 1.
- 4. under a kind of Linux environment according to claim 1 based on user and professional QoS system, it is characterized in that, described multi-queue+LLQ structure is to increase a LLQ formation in three grades of formations of multi-queue, wherein the priority of this LLQ formation is higher than other any formations in three grades of formations, by token bucket is maximum bandwidth reserved of LLQ queue assignment, if network does not take place when congested, the message very first time that enters LLQ is forwarded, and network takes place when congested, the message that enters LLQ then will be by speed limit, message for the overshoot flow will be dropped, when system always preferentially sends message among the LLQ do not have message in LLQ or when surpassing maximum bandwidth reserved for the LLQ configuration, just dispatch the message of other formations in three grades of formations of transmission.
- 5. based on user and professional QoS system, it is characterized in that described DWRR algorithm concrete steps are as follows under a kind of Linux environment according to claim 1:1) for each formation is provided with a difference counter DC, the value of DC is exactly the total amount of byte that scheduler can send at every turn;2) according to the size of formation weights, for each formation is provided with a quantity variable, what variable was represented is the byte traffic volume of distributing to formation;3) scheduler is visited a non-empty queue at every turn, if the message byte number of formation head is greater than DC, scheduler moves to next formation so, if the message byte number of formation head is less than or equal to DC, scheduler is just transmitted this message and the value of DC reduced the byte number that changes message so;4) continue this process, till the byte number of this formation mop message is greater than DC;5) remaining value is accumulated to and uses next time, if formation is empty, then DC is set to 0, and at this moment scheduler will be served next non-empty queue.
- 6. based on user and professional QoS system, it is characterized in that described scheduler is arranged in the formation maintenance module under a kind of Linux environment according to claim 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009101994282A CN102082765A (en) | 2009-11-26 | 2009-11-26 | User and service based QoS (quality of service) system in Linux environment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009101994282A CN102082765A (en) | 2009-11-26 | 2009-11-26 | User and service based QoS (quality of service) system in Linux environment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102082765A true CN102082765A (en) | 2011-06-01 |
Family
ID=44088519
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2009101994282A Pending CN102082765A (en) | 2009-11-26 | 2009-11-26 | User and service based QoS (quality of service) system in Linux environment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102082765A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102333026A (en) * | 2011-06-13 | 2012-01-25 | 中兴通讯股份有限公司 | Message forwarding method and device |
CN102377665A (en) * | 2011-08-12 | 2012-03-14 | 华南理工大学 | IPv6-based radio frequency identification (RFID) router quality of service (QoS) method |
CN102571607A (en) * | 2012-02-28 | 2012-07-11 | 大连梯耐德网络技术有限公司 | Network flow rate distribution equipment |
CN102710506A (en) * | 2012-05-17 | 2012-10-03 | 西安交大捷普网络科技有限公司 | High-performance fluid control method based on linux kernel |
CN103873383A (en) * | 2012-12-12 | 2014-06-18 | 中国移动通信集团上海有限公司 | Queue scheduling method and device |
CN104092626A (en) * | 2014-07-22 | 2014-10-08 | 太仓市同维电子有限公司 | Method for guaranteeing IP converged communication system service quality |
CN104348751A (en) * | 2013-07-31 | 2015-02-11 | 中兴通讯股份有限公司 | Virtual output queue authorization management method and apparatus |
CN105262703A (en) * | 2015-09-25 | 2016-01-20 | 中铁工程装备集团有限公司 | Routing bandwidth allocation method based on double priorities |
CN108471388A (en) * | 2017-02-23 | 2018-08-31 | 中兴通讯股份有限公司 | A kind of method, apparatus that realizing user QoS and base station |
-
2009
- 2009-11-26 CN CN2009101994282A patent/CN102082765A/en active Pending
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012171461A1 (en) * | 2011-06-13 | 2012-12-20 | 中兴通讯股份有限公司 | Method and device for forwarding packet |
CN102333026A (en) * | 2011-06-13 | 2012-01-25 | 中兴通讯股份有限公司 | Message forwarding method and device |
CN102377665A (en) * | 2011-08-12 | 2012-03-14 | 华南理工大学 | IPv6-based radio frequency identification (RFID) router quality of service (QoS) method |
CN102377665B (en) * | 2011-08-12 | 2014-04-02 | 华南理工大学 | IPv6-based radio frequency identification (RFID) router quality of service (QoS) method |
CN102571607B (en) * | 2012-02-28 | 2014-11-05 | 大连梯耐德网络技术有限公司 | Network flow rate distribution equipment |
CN102571607A (en) * | 2012-02-28 | 2012-07-11 | 大连梯耐德网络技术有限公司 | Network flow rate distribution equipment |
CN102710506A (en) * | 2012-05-17 | 2012-10-03 | 西安交大捷普网络科技有限公司 | High-performance fluid control method based on linux kernel |
CN102710506B (en) * | 2012-05-17 | 2015-01-07 | 西安交大捷普网络科技有限公司 | High-performance fluid control method based on linux kernel |
CN103873383A (en) * | 2012-12-12 | 2014-06-18 | 中国移动通信集团上海有限公司 | Queue scheduling method and device |
CN103873383B (en) * | 2012-12-12 | 2017-07-28 | 中国移动通信集团上海有限公司 | A kind of array dispatching method and device |
CN104348751A (en) * | 2013-07-31 | 2015-02-11 | 中兴通讯股份有限公司 | Virtual output queue authorization management method and apparatus |
CN104348751B (en) * | 2013-07-31 | 2019-03-12 | 中兴通讯股份有限公司 | Virtual output queue authorization management method and device |
CN104092626A (en) * | 2014-07-22 | 2014-10-08 | 太仓市同维电子有限公司 | Method for guaranteeing IP converged communication system service quality |
CN105262703A (en) * | 2015-09-25 | 2016-01-20 | 中铁工程装备集团有限公司 | Routing bandwidth allocation method based on double priorities |
CN105262703B (en) * | 2015-09-25 | 2018-06-29 | 中铁工程装备集团有限公司 | A kind of routing bandwidth allocation methods based on dual priority |
CN108471388A (en) * | 2017-02-23 | 2018-08-31 | 中兴通讯股份有限公司 | A kind of method, apparatus that realizing user QoS and base station |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102082765A (en) | User and service based QoS (quality of service) system in Linux environment | |
EP2862301B1 (en) | Multicast to unicast conversion technique | |
US8599870B2 (en) | Channel service manager with priority queuing | |
CN101964758A (en) | Differentiated service-based queue scheduling method | |
CN102158418A (en) | System for content based message processing | |
CN108616458A (en) | The system and method for schedule packet transmissions on client device | |
WO2009130218A1 (en) | A traffic manager and a method for a traffic manager | |
Yang et al. | A new class of priority-based weighted fair scheduling algorithm | |
CN101834787A (en) | Method and system for dispatching data | |
CN107431667A (en) | Packet is dispatched in the network device | |
CN100466593C (en) | Method of implementing integrated queue scheduling for supporting multi service | |
CN105577563B (en) | flow management method | |
CN107896200A (en) | Message scheduling method compatible with virtual link and packet switching mechanism | |
CN101557346B (en) | Polling-based packet queue output scheduling method and packet switching node | |
Yaghmaee et al. | A model for differentiated service support in wireless multimedia sensor networks | |
Tong et al. | Quantum varying deficit round robin scheduling over priority queues | |
Rehman et al. | Comparative Analysis of Scheduling Algorithms in IEEE 802.16 WiMAX | |
Musa et al. | A Comparative Study of Different Queuing Scheduling Disciplines | |
Li et al. | Research and improvement of real-time queue scheduling algorithm | |
Jiwasurat et al. | Hierarchical shaped deficit round-robin scheduling | |
Sharmal et al. | Review of Different Queuing Disciplines in VOIP, Video Conferencing and File Transfer | |
Nandhini | Improved round robin queue management algorithm for elastic and inelastic traffic flows | |
Minagawa et al. | Double WFQ QoS scheduling based on flow number in diffserve network | |
Kim | On guaranteeing the quality of service of conformant traffic in excess bandwidth allocation for shared access networks | |
WO2022022222A1 (en) | Method for sending data packet, and network device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20110601 |