CN100561982C - A kind of high speed exchange arbitration dispatching method based on the fixed length grouping - Google Patents
A kind of high speed exchange arbitration dispatching method based on the fixed length grouping Download PDFInfo
- Publication number
- CN100561982C CN100561982C CNB2007100621740A CN200710062174A CN100561982C CN 100561982 C CN100561982 C CN 100561982C CN B2007100621740 A CNB2007100621740 A CN B2007100621740A CN 200710062174 A CN200710062174 A CN 200710062174A CN 100561982 C CN100561982 C CN 100561982C
- Authority
- CN
- China
- Prior art keywords
- output port
- input port
- port
- scheduling
- matrix
- 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.)
- Active
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a kind of high speed exchange arbitration dispatching method based on the fixed length grouping, it relates to the Data Interchange Technology in communication system IP exchange and the ATM exchange field.The present invention adopts the input buffering pattern of virtual output work queue, is a kind of related pointer of the iteration repeatedly iterative cycles priority matching process that slides, and after it finished the input port request of data, the scheduling of data and arbitration provided scheduling result.The present invention has fast convergence rate, and fairness is good, and it is few to take hardware resource, is easy to characteristics such as hardware realization, is particularly suitable for being applied to communication system high speed IP switch and atm switching apparatus.
Description
Technical field
The present invention relates to communication system IP exchange and arbitrate dispatching method, be particularly suitable for being applied to communication system high speed IP switch and atm switching apparatus with a kind of high speed exchange in the ATM exchange field based on the fixed length grouping.
Background technology
Along with the development of the communication technology and electronic technology, high-end router atm switching apparatus exchange capacity is increasing at present, and port speed also rapidly increases.Increasing function begins to realize with hardware mode, is embodied in ASIC (dedicated IC chip), FPGA and uses more and more widely; Switching equipment is abandoned gradually using shared bus and is used switching backplane, promptly begins generally to adopt the switch type route technology; Under this environment, the empty Crossbar type switching fabric that divides is used in high-end router gradually widely.The empty Crossbar of branch type switching fabric is introduced scheduling processing method and is solved the competition of each port to resource, because its important effect, scheduling processing method has become the research emphasis of Crossbar type switching fabric.
Summary of the invention
Technical problem to be solved by this invention be to avoid the weak point in the above-mentioned background technology and provide a kind of fixed length packet high-speed exchange arbitration dispatching method that changes based on the related pointer wheel of iteration, the present invention have fast convergence rate, hardware designs simple, take that hardware resource is few, the scheduling fairness is good, be easy to characteristics such as priority design.
The object of the present invention is achieved like this, comprises step:
1. request process, a plurality of data buffering formations are arranged, in request stage, if when data buffering formation non-NULL promptly has cell to export on each input port, this input port just proposes the transmission request to corresponding output port, and the transmission request of all input ports forms request matrix R;
2. licensing process, do not set up the idle output port of sendaisle as yet with input port, after the request of receiving, relation according to scheduling pointer and input port and output port, according to priority orders that sets in advance and scheduling pointer position, select an input port to authorize from input port, all output ports form authorization matrix G to the Authorization result of input port;
3. reception process, after input port is obtained the authorization, relation according to scheduling pointer and input port and output port, according to priority orders that sets in advance and scheduling pointer position, from the output port of authorizing, accept an output port, the input port that this is obtained the authorization and the output port of acceptance are finished coupling, and notify this output port, and all input ports are accepted matrix A to result's formation of output port mandate;
4. after each coupling, appropriate ports do not carry out the coupling of next round as yet, formation is accepted after the matrix A according to the result who accepts matrix A, produce new request matrix R, repetitive operation 1., 2., 3. step, handle convergence up to this time, convergence is meant the 1., 2., 3. in the step, do not produce new coupling;
5. after each processing convergence, the scheduling pointer position changes, each input port changes to the priority of the output port of the reception process correspondence of output port, output port changes to the priority of the licensing process corresponding input end mouth of input port, obtain the justice of arbitration scheduling, finish high speed exchange arbitration scheduling based on the fixed length grouping;
Above step 2. in, the priority orders that sets in advance is meant (1) on output port, scheduling pointer residing position input port processed priority orders is a limit priority; (2) port handled at first of output port is the input port at scheduling pointer place, gives authorisation process to the data of request matrix R according to from top to bottom priority orders thereafter; (3) request of any one input port is on all output ports, and its priority that is authorized to has nothing in common with each other; The scheduling pointer position is meant that the scheduling pointer position remains on the diagonal of request matrix R;
Above step 3. in, the priority orders that sets in advance is meant (1) on input port, scheduling pointer residing position output port processed priority orders is a limit priority; (2) port handled at first of input port is accepted processing to the data of authorization matrix G according to priority orders from right to left thereafter for the output port at scheduling pointer place.(3) mandate of any one output port is on all input ports, and its received priority has nothing in common with each other; The scheduling pointer position is meant the scheduling pointer position, and it remains on the diagonal of authorization matrix G;
The present invention compares with background technology, has following advantage:
1) the present invention adopts related scheduling pointer, makes that this method iterations is few, and fast convergence rate has solved the problem that bandwidth availability ratio reduces under the high load condition.
Limit iterations of the present invention has only (N is the switching port number) N/2 time, and the iterations of this method has only (log under the smooth traffic
2N)/2 time, convergence rate increases greatly.Simultaneously because the scheduling pointer of this method is pre-designed, it is related that input port is accepted with authorizing of output port, the present invention adopts one group of overall scheduling pointer processing mode, each port passes through different phase associations to the scheduling pointer, the authorisation process that has so just guaranteed the acceptance processing of any one input port or output port can be not overlapping, not only make background technology be solved in the problem that high load condition lower port utilance descends, simultaneously because after having made full use of the each iteration of related scheduling pointer, input/output port is handled the still uncompetitive principle of limit priority, and in the process of iteration, be applied, make maximum iteration time be reduced to N/2 time, optimized algorithm greatly, fairness is also strengthened greatly.
2) hardware complexity of the present invention lowers greatly, and hardware designs is simple, it is few to take hardware resource.
The present invention authorizes in input acceptance and output, needs 2N priority encoder to handle the process of the request of input port or output port, mandate, acceptance, simplicity of design; In the pointer processing, the present invention has only one group of association pointer, and pointer is controlled simply once dispatching the processing back from the driving wheel commentaries on classics.
3) the present invention is good to the scheduling fairness of data processing.
The present invention guarantees in N cell time-slot, the chance that the team of each an input port cell must obtain once to send, and therefore the fairness that data are handled is than the N in the background technology
2Individual cell time-slot obtains once to authorize and obviously improves.
Description of drawings
When Fig. 1 is the invention process 4 * 4 exchanges, the schematic diagram of an iteration.
When Fig. 2 was the invention process 4 * 4 exchanges, pointer circulation wheel changeed processing schematic diagram once.
Embodiment
With reference to Fig. 1, Fig. 2, when Fig. 1 is the invention process 4 * 4 exchanges, the schematic diagram of an iteration; When Fig. 2 was the invention process 4 * 4 exchanges, pointer circulation wheel changeed processing schematic diagram once.The inventive method comprises step:
1. request process, a plurality of data buffering formations are arranged, in request stage, if when data buffering formation non-NULL promptly has cell to export on each input port, this input port just proposes the transmission request to corresponding output port, and the transmission request of all input ports forms request matrix R.Embodiment is shown in R matrix among Fig. 1.
2. licensing process, do not set up the idle output port of sendaisle as yet with input port, after the request of receiving, relation according to scheduling pointer and input/output port, according to priority orders that sets in advance and scheduling pointer position, select an input port to authorize from input port, the Authorization result of all output ports forms authorization matrix G.Embodiment is shown in G matrix among Fig. 1.The present invention is in licensing process, and it remains at the scheduling pointer position on the diagonal of request matrix R.The present invention is at licensing process, and the input port that output port is handled at first is the input port at scheduling pointer place, the data of request matrix R is handled giving authorisation process thereafter according to priority orders from top to bottom.The present invention is in licensing process, and the request of any one input port is on all output ports, and its priority orders that is authorized to has nothing in common with each other.
3. reception process, after input port is obtained the authorization, relation according to scheduling pointer and input port, according to priority orders that sets in advance and scheduling pointer position, from the output port of authorizing, accept an output port, this input port and output port are finished coupling, and notify this output port, and matrix A is accepted in the result's formation of accepting of all input ports.Embodiment is shown in A matrix among Fig. 1.The present invention dispatches pointer position in reception process, and it remains on the diagonal of authorization matrix G, and all inputs are accepted and the output mandate all is to be set to limit priority at scheduling pointer position place; The present invention dispatches pointer position in reception process, and the mandate of any one output port is on all input ports, and its received priority has nothing in common with each other.
4. after each coupling, appropriate ports do not carry out the coupling of next round as yet, transmit passage in case set up, will in the iteration of back, be cancelled, form and to accept after the matrix A to produce new request matrix R according to the result who accepts matrix A, repetitive operation 1., 2., 3. step, handle convergence up to this time, convergence is meant the 1., 2., 3. in the step, do not produce new coupling;
5. after each processing convergence, the scheduling pointer changes, embodiment as shown in Figure 2, after pointer changes, the priority that corresponding output port is handled in the acceptance of each input port changes, the priority of corresponding input end mouth changes during the authorisation process of output port, obtains the justice of arbitration scheduling, finishes the high speed exchange arbitration scheduling based on the fixed length grouping.
Claims (1)
1. dispatching method is arbitrated in the high speed exchange based on the fixed length grouping, it is characterized in that comprising step:
1. request process, a plurality of data buffering formations are arranged, in request stage, if when data buffering formation non-NULL promptly has cell to export on each input port, this input port just proposes the transmission request to corresponding output port, and the transmission request of all input ports forms request matrix R;
2. licensing process, do not set up the idle output port of sendaisle as yet with input port, after the request of receiving, relation according to scheduling pointer and input port and output port, according to priority orders that sets in advance and scheduling pointer position, select an input port to authorize from input port, all output ports form authorization matrix G to the Authorization result of input port;
3. reception process, after input port is obtained the authorization, relation according to scheduling pointer and input port and output port, according to priority orders that sets in advance and scheduling pointer position, from the output port of authorizing, accept an output port, the input port that this is obtained the authorization and the output port of acceptance are finished coupling, and notify this output port, and all input ports are accepted matrix A to result's formation of output port mandate;
4. after each coupling, appropriate ports do not carry out the coupling of next round as yet, formation is accepted after the matrix A according to the result who accepts matrix A, produce new request matrix R, repetitive operation 1., 2., 3. step, handle convergence up to this time, convergence is meant the 1., 2., 3. in the step, do not produce new coupling;
5. after each processing convergence, the scheduling pointer position changes, each input port changes to the priority of the output port of the reception process correspondence of output port, output port changes to the priority of the licensing process corresponding input end mouth of input port, obtain the justice of arbitration scheduling, finish high speed exchange arbitration scheduling based on the fixed length grouping;
Above step 2. in, the priority orders that sets in advance is meant (1) on output port, scheduling pointer residing position input port processed priority orders is a limit priority; (2) port handled at first of output port is the input port at scheduling pointer place, gives authorisation process to the data of request matrix R according to from top to bottom priority orders thereafter; (3) request of any one input port is on all output ports, and its priority that is authorized to has nothing in common with each other; The scheduling pointer position is meant that the scheduling pointer position remains on the diagonal of request matrix R;
Above step 3. in, the priority orders that sets in advance is meant (1) on input port, scheduling pointer residing position output port processed priority orders is a limit priority; (2) port handled at first of input port is accepted processing to the data of authorization matrix G according to priority orders from right to left thereafter for the output port at scheduling pointer place
(3) mandate of any one output port is on all input ports, and its received priority has nothing in common with each other; The scheduling pointer position is meant the scheduling pointer position, and it remains on the diagonal of authorization matrix G
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007100621740A CN100561982C (en) | 2007-06-18 | 2007-06-18 | A kind of high speed exchange arbitration dispatching method based on the fixed length grouping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007100621740A CN100561982C (en) | 2007-06-18 | 2007-06-18 | A kind of high speed exchange arbitration dispatching method based on the fixed length grouping |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101072175A CN101072175A (en) | 2007-11-14 |
| CN100561982C true CN100561982C (en) | 2009-11-18 |
Family
ID=38899170
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2007100621740A Active CN100561982C (en) | 2007-06-18 | 2007-06-18 | A kind of high speed exchange arbitration dispatching method based on the fixed length grouping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100561982C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108173787A (en) * | 2017-11-30 | 2018-06-15 | 成都成电光信科技股份有限公司 | Data transmission path selection method, Method of Scheduling Parallel and the device of FC interchangers |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101989942B (en) * | 2009-08-07 | 2012-10-31 | 无锡江南计算技术研究所 | Arbitration control method, communication method, arbitrator and communication system |
| CN101626347B (en) * | 2009-08-13 | 2012-07-04 | 华为技术有限公司 | Cell-switching method, and switching equipment and switching system |
| CN101795201B (en) * | 2010-01-28 | 2015-04-15 | 中国电子科技集团公司第五十四研究所 | Multi-stage survivable switching structure device |
| CN107743101B (en) * | 2017-09-26 | 2020-10-09 | 杭州迪普科技股份有限公司 | Data forwarding method and device |
| CN113110943B (en) * | 2021-03-31 | 2023-04-25 | 中国人民解放军战略支援部队信息工程大学 | Software defined switching structure and data switching method based on same |
-
2007
- 2007-06-18 CN CNB2007100621740A patent/CN100561982C/en active Active
Non-Patent Citations (4)
| Title |
|---|
| 共享缓冲在ATM 交换矩阵中的研究与应用. 李文江.无线电工程，第36卷第2期. 2006 |
| 共享缓冲在ATM 交换矩阵中的研究与应用. 李文江.无线电工程，第36卷第2期. 2006 * |
| 基于VOQ 输入缓存交换系统调度算法研究. 赵增辉等.无线电通信技术，第32卷第6期. 2006 |
| 基于VOQ 输入缓存交换系统调度算法研究. 赵增辉等.无线电通信技术，第32卷第6期. 2006 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108173787A (en) * | 2017-11-30 | 2018-06-15 | 成都成电光信科技股份有限公司 | Data transmission path selection method, Method of Scheduling Parallel and the device of FC interchangers |
| CN108173787B (en) * | 2017-11-30 | 2020-12-18 | 成都成电光信科技股份有限公司 | Data transmission path selection method, parallel scheduling method and device for FC switch |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101072175A (en) | 2007-11-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100561982C (en) | A kind of high speed exchange arbitration dispatching method based on the fixed length grouping | |
| US20020176431A1 (en) | Multiserver scheduling system and method for a fast switching element | |
| US20150063348A1 (en) | Implementing hierarchical high radix switch with timesliced crossbar | |
| CN1214168A (en) | Efficient output-request packet switch and method | |
| CN108173787B (en) | Data transmission path selection method, parallel scheduling method and device for FC switch | |
| CN104486221B (en) | A kind of double access infrastructures based on two-dimensional grid network-on-chip | |
| CN1819523A (en) | Parallel interchanging switching designing method | |
| US7408947B2 (en) | Method and apparatus for scheduling packets and/or cells | |
| CN1449160A (en) | Packet dispatch for system containing non-blocking exchange structure and line connecting interface | |
| CN101964747B (en) | Two-stage exchanging structure working method based on preposed feedback | |
| US20070133585A1 (en) | Method and device for scheduling interconnections in an interconnecting fabric | |
| US7912068B2 (en) | Low-latency scheduling in large switches | |
| CN1878132A (en) | Method and system for realizing variable-length packet switching | |
| CN101263680B (en) | Method and device for scheduling unicast and multicast traffic in an interconnecting fabric | |
| Xiaopeng et al. | Round robin arbiters for virtual channel router | |
| US20050163127A1 (en) | Buffer switch and scheduling method thereof | |
| CN109327405A (en) | Message order-preserving method and the network equipment | |
| CN111464461B (en) | Priority scheduling method and device for switch | |
| Jiang et al. | A 2-stage matching scheduler for a VOQ packet switch architecture | |
| Pan et al. | Pipelined two step iterative matching algorithms for CIOQ crossbar switches | |
| Qu et al. | Contention‐tolerant crossbar packet switches | |
| CN101626347A (en) | Cell-switching method, and switching equipment and switching system | |
| Gao et al. | Distributed weight matching dispatching scheme in MSM Clos-network packet switches | |
| Delgado-Frias et al. | A VLSI crossbar switch with wrapped wave front arbitration | |
| Chrysos et al. | Practical high-throughput crossbar scheduling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |