CN106453137A - Dynamic multicast enqueue algorithm for relieving HOL Blocking - Google Patents

Dynamic multicast enqueue algorithm for relieving HOL Blocking Download PDF

Info

Publication number
CN106453137A
CN106453137A CN201610833616.6A CN201610833616A CN106453137A CN 106453137 A CN106453137 A CN 106453137A CN 201610833616 A CN201610833616 A CN 201610833616A CN 106453137 A CN106453137 A CN 106453137A
Authority
CN
China
Prior art keywords
multicast
packet
queue
algorithm
team
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201610833616.6A
Other languages
Chinese (zh)
Other versions
CN106453137B (en
Inventor
熊庆旭
梁佳诚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beihang University
Original Assignee
Beihang University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beihang University filed Critical Beihang University
Priority to CN201610833616.6A priority Critical patent/CN106453137B/en
Publication of CN106453137A publication Critical patent/CN106453137A/en
Application granted granted Critical
Publication of CN106453137B publication Critical patent/CN106453137B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/50—Queue scheduling
    • H04L47/62—Queue scheduling characterised by scheduling criteria
    • H04L47/625—Queue scheduling characterised by scheduling criteria for service slots or service orders
    • H04L47/6255—Queue scheduling characterised by scheduling criteria for service slots or service orders queue load conditions, e.g. longest queue first
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/20—Support for services
    • H04L49/201—Multicast operation; Broadcast operation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/30—Peripheral units, e.g. input or output ports
    • H04L49/3018—Input queuing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/30—Peripheral units, e.g. input or output ports
    • H04L49/3027—Output queuing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/30—Peripheral units, e.g. input or output ports
    • H04L49/3045—Virtual queuing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/30—Peripheral units, e.g. input or output ports
    • H04L49/3072—Packet splitting

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses a dynamic multicast enqueue algorithm for relieving HOL Blocking, and belongs to the technical field of packet scheduling of unicast and multicast mixed business in a CICQ structure packet switch. The purpose of the invention is to relieve the multicast HOL Blocking problem in the packet scheduling of the CICQ switch. The dynamic multicast enqueue algorithm disclosed by the invention comprises the following steps: dividing two types of HOL Blocking, which are separately HOL Blocking based on transmission requirements and HOL Blocking based on Work-Conserving; and considering the HOL Blocking based on Work-Conserving at first and then considering the HOL Blocking based on transmission requirements to propose a new multicast enqueue solution MDA algorithm according to the relationship between the arrival multicast packet direction and a current cross caching state of the CICQ switch. Compared with the typical multicast enqueue algorithm, the algorithm provided by the invention has a lower average packet delay.

Description

A kind of joined the team algorithm with alleviating the dynamic multicast as target for the HOL Blocking
Technical field
The invention belongs to packet scheduling technical field under single multicast mixed service in CICQ structure packet switch.
Background technology
Output queueing packet switch has the demand of N times of speed-up ratio, and N is the number of input port, therefore inapplicable In high-speed high capacity network.Input queue's structure no speed-up ratio requires, but compared with Output queueing, input queue's structure is deposited It is grouped blocking, i.e. HOL (Head of Line) Blocking problem in head.HOL Blocking refers to when same in input port The packet whereabouts queued up in queue is not simultaneously as head packet there is no transmission opportunity so that head is grouped packet quilt below Blocking, thus losing the chance that can transmit, result makes switch percent of pass decline.
When only existing unicast service, N number of VOQ is set in each input port, can thoroughly solve HOL Blocking problem.However, under the conditions of single multicast mixed service, for the CICQ (Combined of a N × N-port Input and Crossbar Queued) fabric switch, reach the whereabouts being grouped and be up to 2N-1 kind.This means that thorough Bottom solves the problems, such as HOL Blocking, needs to arrange 2N-1 multicast virtual output queue, this is in actual large capacity exchanger Unpractical.For alleviating HOL Blocking problem, existing mainstream technology scheme is to arrange k (1 in input port for multicast service <k<<2N- 1) individual multicast virtual output queue, then according to multicastapackets are joined the team corresponding multicast by certain rule of joining the team VOQ.In general, multicast join the team algorithm should so that in multicast virtual output queue after joining the team packet as far as possible full Sufficient three below rule：1) the head packet diversity of k queue is good, can cover all of N number of output port as far as possible；2) whereabouts Same or like multicastapackets are included into same queue；3) different queue load is tried one's best and is equalized.
The existing multicast method of joining the team adopts static policies, and such as Modulo algorithm passes through to calculate the fan reaching multicastapackets Go out several results that multicast queue number k modulus are obtained, to determine the queue of joining the team of this multicastapackets；Vector algorithm is by for often Individual multicast queue arranges corresponding characteristic vector, calculates the distance of the whereabouts vector reaching packet and each characteristic vector, according to The distance calculating is determining the queue of joining the team of this multicastapackets.But, existing multicast is joined the team, and strategy is all uncombined to reach multicast Relation between packet whereabouts and current CICQ switch buffer status, does dynamic adjustment to rule of joining the team.
The present invention proposes a kind of new multicast dynamic algorithm MDA (Multicast Dynamic Algorithm), this calculation Method is based on current time slots CICQ switch buffer status, does dynamic adjustment to the whereabouts of the multicastapackets that this time slot reaches.
Content of the invention
It is an object of the invention to provide alleviating HOL Blocking problem in CICQ structure to reduce packeting average time delay Multicastapackets are joined the team algorithm.
For achieving the above object, the technology path of present invention employing is：
First, divide the type of two kinds of input port multicast HOL Blocking in CICQ switching fabric：
1. the HOL Blocking in existing packet scheduling technology refers to, for a certain input port, its all multicast is empty Intend the intersection caching corresponding to the output port that the packet of output queue head is gone to and all can not receive new packet, and at least one The intersection caching corresponding to output port that the secondary packet of queue is gone to can receive new packet, and the present invention is referred to as to be based on transmission The HOL Blocking of demand；
2. the present invention works in the angle of Work-Conserving from switch, proposes one kind and is based on Work- The HOL Blocking of Conserving.That is it is all space-time when the corresponding row of a certain output port intersect caching, certain is defeated In inbound port, the head packet whereabouts of all VOQs does not comprise this output port, and the secondary packet of an at least queue Whereabouts comprises the phenomenon of this output port, then claim this input port that the HOL Blocking based on Work-Conserving occurs.
Then, the division according to HOL Blocking type, determines the rule of joining the team of packet.The present invention considers to be based on first The HOL Blocking of Work-Conserving, secondly considers the HOL Blocking based on transmission demand.For current time slots Reach the multicastapackets of input port i,
If 1. in the corresponding all of intersection caching of a certain output port of current time slots, total number packets are less than or equal to 1, and arrive The whereabouts reaching packet comprises this output port, then the virtual multicast output team the shortest by being grouped length in the input port i that joins the team Row；
If input port i transmit its arbitrary virtual queue k head be grouped into corresponding intersect caching after, input port i Corresponding intersect caching and still remove the head of the other virtual queues outside virtual queue k in receivability input port i be grouped, then reach Packet is joined the team according to Vector algorithm；Otherwise, the virtual multicast output team the shortest by being grouped length in the input port i that joins the team Row.
Beneficial effects of the present invention：The invention provides a kind of new based on current time slots CICQ switch buffer status and The dynamic multicast of dispatching requirement is joined the team algorithm, has the average packet time delay lower than existing main flow algorithm.
Brief description
Fig. 1 is combined input crossbar contact caching queuing (CICQ) fabric switch block diagram；
Fig. 2 is the technical thought of MDA algorithm of the present invention.
Specific embodiment
Fig. 1 gives the CICQ block diagram of structure packet switch.Under the conditions of single multicast mixed service, multicast service reaches When, switch scheduling process is divided into：Multicast is joined the team>Input scheduling>Output scheduling three step, inventive algorithm consideration group is sown into Team's stage.
When multicastapackets reach input port i, multicast virtual queue is selected to join the team by the following step.
1st step calculates the length of the current the shortest multicast queue of input port i.
If the shortest multicast queue length of the 2nd step is more than 1, join the team according to Vector algorithm；Otherwise enter the 3rd step.
If the 3rd step reaches certain output port corresponding intersection caching be up to non-NULL that packet is gone to, this packet Entering input port i queue length multicast queue the shortest, if there is the shortest queue of multiple length identicals, selecting head packet The maximum queue of weight is joined the team, and head packet weighing computation method is：0.5 × this packet reach switch when fanout square/ The current fanout of this packet × this multicastapackets stand-by period；Otherwise, enter the 4th step.
If input port i has packet can adopt to intersecting caching, this packet in next slot transmission after the 4th step input scheduling With entering queue length multicast queue the shortest with the 3rd step identical method；Otherwise join the team according to Vector algorithm.

Claims (1)

1. a kind of joined the team algorithm with alleviating the dynamic multicast as target for the HOL Blocking
Vector algorithm is that a kind of existing multicast is joined the team scheme, and its algorithmic procedure is：From k feature of N-dimensional Peug Zixian Vector v1,v2,…,vkCorrespond to k multicast queue respectively, the value of each element of each characteristic vector is 0 or 1, and each two is special Levy vector mutually orthogonal；
For the multicastapackets reaching, define whereabouts vector D according to its whereaboutsa；If comprising output port j to this packet whereabouts, Da[j]=1, otherwise Da[j]=0.In v1,v2,…,vkIn find out and DaDistance characteristic vector the shortest, then reach packet and enter The corresponding multicast queue of this feature vector；
The present invention provide a kind of joined the team algorithm with alleviating the dynamic multicast as target for the HOL Blocking, it is characterized in that：
When input port i has multicastapackets to reach, multicast virtual queue is selected to join the team as follows：
1st step calculates the length of the current the shortest multicast queue of input port i；
If the shortest multicast queue length of the 2nd step is more than 1, join the team according to Vector algorithm；Otherwise enter the 3rd step；
If the 3rd step reaches certain output port corresponding intersection caching be up to non-NULL that packet is gone to, this packet enters Input port i queue length multicast queue the shortest, if there is the shortest queue of multiple length identicals, selects head packet weight Maximum queue is joined the team, and head packet weighing computation method is：0.5 × this packet reach switch when fanout square/this point Organize current fanout × this multicastapackets stand-by period；Otherwise, enter the 4th step；
If input port i has packet can adopt to intersecting caching, this packet in next slot transmission after the completion of the 4th step input scheduling With entering queue length multicast queue the shortest with the 3rd step identical method；Otherwise join the team according to Vector algorithm.
CN201610833616.6A 2016-09-20 2016-09-20 It is a kind of to be joined the team method using alleviating HOL Blocking as the dynamic multicast of target Active CN106453137B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610833616.6A CN106453137B (en) 2016-09-20 2016-09-20 It is a kind of to be joined the team method using alleviating HOL Blocking as the dynamic multicast of target

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610833616.6A CN106453137B (en) 2016-09-20 2016-09-20 It is a kind of to be joined the team method using alleviating HOL Blocking as the dynamic multicast of target

Publications (2)

Publication Number Publication Date
CN106453137A true CN106453137A (en) 2017-02-22
CN106453137B CN106453137B (en) 2019-04-05

Family

ID=58167054

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610833616.6A Active CN106453137B (en) 2016-09-20 2016-09-20 It is a kind of to be joined the team method using alleviating HOL Blocking as the dynamic multicast of target

Country Status (1)

Country Link
CN (1) CN106453137B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110858791A (en) * 2018-08-22 2020-03-03 华为技术有限公司 Distributed parallel transmission method, device, equipment and storage medium
CN115941626A (en) * 2022-12-06 2023-04-07 北京航空航天大学 Acceleration ratio-free work keeping packet scheduling solution for packet switch

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101286866A (en) * 2008-05-30 2008-10-15 杭州华三通信技术有限公司 Multicast implementing method and system based on switching network of high-speed peripheral extended interface
CN101478483A (en) * 2009-01-08 2009-07-08 中国人民解放军信息工程大学 Method for implementing packet scheduling in switch equipment and switch equipment
CN101656659A (en) * 2008-08-19 2010-02-24 中兴通讯股份有限公司 Method for caching mixed service flow and method and device for storing and forwarding mixed service flow
CN101695052A (en) * 2009-10-21 2010-04-14 中国人民解放军理工大学 Small cross point buffer high-property crossbar dispatching method
CN103384225A (en) * 2013-03-15 2013-11-06 北京安拓思科技有限责任公司 Flow control method based on input intersection cache fast packet switched network
CN103581056A (en) * 2013-10-25 2014-02-12 盛科网络（苏州）有限公司 Method for achieving congestion management in network communication through layered queue scheduling
CN105429898A (en) * 2015-11-02 2016-03-23 北京航空航天大学 CICQ structure crossbuffer queue balance packet scheduling algorithm

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101286866A (en) * 2008-05-30 2008-10-15 杭州华三通信技术有限公司 Multicast implementing method and system based on switching network of high-speed peripheral extended interface
CN101656659A (en) * 2008-08-19 2010-02-24 中兴通讯股份有限公司 Method for caching mixed service flow and method and device for storing and forwarding mixed service flow
CN101478483A (en) * 2009-01-08 2009-07-08 中国人民解放军信息工程大学 Method for implementing packet scheduling in switch equipment and switch equipment
CN101695052A (en) * 2009-10-21 2010-04-14 中国人民解放军理工大学 Small cross point buffer high-property crossbar dispatching method
CN103384225A (en) * 2013-03-15 2013-11-06 北京安拓思科技有限责任公司 Flow control method based on input intersection cache fast packet switched network
CN103581056A (en) * 2013-10-25 2014-02-12 盛科网络（苏州）有限公司 Method for achieving congestion management in network communication through layered queue scheduling
CN105429898A (en) * 2015-11-02 2016-03-23 北京航空航天大学 CICQ structure crossbuffer queue balance packet scheduling algorithm

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
XIONG Q X.: "Research on packet scheduling in input-queued switches", 《JORNAL ON COMMUNICATIONS 2005》 *
张元昊，熊庆旭: "CICQ结构中逼近work-conserving的分组调度算法", 《北京航空航天大学学报》 *
梁佳诚，熊庆旭，闫付龙，萧翰，张元昊: "基于Work-Conserving的CICQ结构中单组波分组调度算法", 《北京航空航天大学学报》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110858791A (en) * 2018-08-22 2020-03-03 华为技术有限公司 Distributed parallel transmission method, device, equipment and storage medium
CN115941626A (en) * 2022-12-06 2023-04-07 北京航空航天大学 Acceleration ratio-free work keeping packet scheduling solution for packet switch

Also Published As

Publication number Publication date
CN106453137B (en) 2019-04-05

Similar Documents

Publication Publication Date Title
CN104486237B (en) Without out-of-order packet route and dispatching method in clos networks
JP5034491B2 (en) Priority bandwidth control method in switch
Xue et al. Dart: Divide and specialize for fast response to congestion in RDMA-based datacenter networks
CN101695051A (en) Queue length balance dispatching method used for buffered Crossbar
CN110351187A (en) Data center network Road diameter switches the adaptive load-balancing method of granularity
Bianco et al. On the number of input queues to efficiently support multicast traffic in input queued switches
CN106453137A (en) Dynamic multicast enqueue algorithm for relieving HOL Blocking
CN105429898B (en) A kind of CICQ structures intersect the balanced packet scheduling algorithm of buffer queue
Chrysos et al. Discharging the network from its flow control headaches: Packet drops and hol blocking
Kim et al. Performance analysis of the multiple input-queued packet switch with the restricted rule
CN106453134B (en) A kind of CICQ fabric switch grouping scheduling method for coordinating single multicast competition based on virtual queue length
CN105072046A (en) Delay tolerant network congestion prevention method based on data concurrence and forwarding by token control node
Yu et al. Multi-level round-robin multicast scheduling with look-ahead mechanism
Chrysos Congestion management for non-blocking Clos networks
Liao et al. Longest queue first in round-robin matching for input-queued switches
Dong et al. Long round-trip time support with shared-memory crosspoint buffered packet switch
Yoshigoe The crosspoint-queued switches with virtual crosspoint queueing
Hassen et al. A scalable packet-switch based on output-queued nocs for data centre networks
Gao An adaptive scheduling algorithm for multi-priority traffic in load-balanced switch
Radusinovic et al. Analysis of WRR scheduling algorithm frame size impact on CQ switch performance
Pan et al. CQPPS: A scalable multi‐path switch fabric without back pressure
Yeung et al. A novel feedback mechanism for load balanced two-stage switches
Zheng et al. Analysis and implementation of a QoS optimization method for access networks
Chrysos et al. Distributed WFQ scheduling converging to weighted max–min fairness
Skalis et al. Performance guarantees in partially buffered crossbar switches

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant