CN101631067B - Method, device and system for packet switching - Google Patents
Method, device and system for packet switching Download PDFInfo
- Publication number
- CN101631067B CN101631067B CN2008101166314A CN200810116631A CN101631067B CN 101631067 B CN101631067 B CN 101631067B CN 2008101166314 A CN2008101166314 A CN 2008101166314A CN 200810116631 A CN200810116631 A CN 200810116631A CN 101631067 B CN101631067 B CN 101631067B
- Authority
- CN
- China
- Prior art keywords
- exchange
- module
- switching
- request
- switching module
- 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 method, a device and a system for packet switching, and belongs to the field of packet switching. The method comprises the following steps: receiving an original switching request sent by an uplink switching module; copying the original switching request to obtain a plurality of switching requests, arbitrating the switching requests according to the states of ports connected to a vertical-horizontal cross switching module and a downlink switching module, outputting the obtained arbitration result to the uplink switching module; receiving cells sent by the uplink switching module; and according to the arbitration result, forwarding the cells to the downlink switching module. The device comprises an input request processing module, an arbitration module and an output permission processing module. The method, the device and the system have the advantages of realizing the intercommunication between the packets, expanding the number of ports of the packet switching system, enhancing the switching capacity, simultaneously being compatible with the original SM system, and reducing the cost of the switching system, along with strong expandability.
Description
Technical field
The present invention relates to field of packet switching, particularly a kind of grouping exchange method, device and system.
Background technology
Along with development of internet technology, the exchange capacity demand of exchange chip is increased day by day, to the exchange chip application flexibility, many scenes are used that higher demand has also been arranged.The exchange capacity that improves exchange chip in the prior art generally has 2 kinds of approach, and first kind is the speed that improves port, reaches higher exchange capacity; Second kind is the port number that increases exchange chip, reaches higher exchange capacity.
When the mode that adopt to improve port speed improves exchange capacity, cause network chip and exchange chip incompatible, need the port speed of the speed raising exchange chip of chip Network Based, limited the raising of exchange capacity through regular meeting.Therefore, adopt the mode that increases the switching chip port number to improve exchange capacity in the prior art usually.
The structural representation of the big capacity fixed length grouping cell switching net that provides referring to Fig. 1, what this fixed length grouping cell switching host will adopt is three-level structure, wherein the first order and the third level are same shared buffer memory exchange chip (Switching Memory chip, SM), the first order is expressed as SMi---up Switching Module, the third level is expressed as SMe---and descending Switching Module is used to finish the cache exchanging and the scheduling of cell; The second level be in length and breadth the cross exchange chip (Switching Crossbar chip SC), is used to finish the arbitration of cell request, realizes the intercommunication between the multi-disc SM.Cell is buffered among the SMi, during cell switching, proposes the exchange request by SMi to SC, and this request is organized in bitmap (bit) mode, i.e. 1 outbound port of the corresponding SC of each bit.In the SC switching network, a plurality of inputs may be competed an output simultaneously, be to have competition between the input and output, SC need arbitrate the exchange request according to arbitration algorithm, after arbitration is passed through, SC returns the allowance signal to SMi, SMi accesses cell come, and gives SC after receiving and permitting signal, SC exchanges it, outputs to purpose SMe from corresponding output port.
SC internal condition application scenarios can be divided into N group (group), between 1 group and 1 SMi 1 link can only be arranged.Group can be 1 exchange chip, also can be the grouping of exchange chip internal logic.The inner connection that can realize being input to any output of group can not be striden the group exchange between the group.
In realizing process of the present invention, the inventor finds that there is following problem at least in prior art in the prior art:
When improving exchange capacity by increasing the SC port number, the bit wide of SMi cell request will be increased to consistent with SC outbound port number accordingly, for this reason, need to revise the form of SM cell, cause the expense of cell to increase, can not compatiblely have the SM chip simultaneously, influence Products Development;
When the SM chip capacity that is connected to SC is unequal, in order to realize linear speed intercommunication between the SM, need to increase be connected between low capacity SM and the SC port number, the port number that heap(ed) capacity SM is connected with SC in port number that SM all in the system is connected with SC and the system is equated, increase the power consumption and the area of SM chip, and then increased the cost of chip.
Summary of the invention
Exchange capacity during for raising packet switching on the basis of compatible original system, the embodiment of the invention provides a kind of grouping exchange method, device and system.Described technical scheme is as follows:
A kind of grouping exchange method is used for cross exchange module in length and breadth, and to realize the exchange to the information in the shared buffer memory Switching Module, described shared buffer memory Switching Module comprises up Switching Module and descending Switching Module, and described method comprises:
Receive the original switching request that described up Switching Module sends;
Bit wide according to described original switching request, and the port number of the described module of cross exchange in length and breadth duplicates described original switching request, makes total bit wide of many parts of exchange requests that obtain after duplicating equal the output port number of the described module of cross exchange in length and breadth;
Whether the output port of checking each grouping of the described module of cross exchange in length and breadth links to each other with descending Switching Module;
Result according to checking masks the exchange request of going to the output port that does not connect descending Switching Module;
Exchange request after the shielding processing is arbitrated,, the arbitration result that obtains is exported to described up Switching Module to select a switching path in the described many parts of exchange requests as cell;
Receive the cell that described up Switching Module sends; Result according to described arbitration is transmitted to described descending Switching Module with described cell.
A kind of packet switching system, described system comprises: shared buffer memory Switching Module and cross exchange module in length and breadth;
Described shared buffer memory Switching Module comprises:
Up Switching Module is used for sending the original switching request to the described module of cross exchange in length and breadth, and after receiving the allowance signal of described original switching request, dispatches the concurrent unit of delivering letters and give the described module of cross exchange in length and breadth;
Descending Switching Module is used to receive the cell of the described module forwards of cross exchange in length and breadth, and described cell is sent to corresponding equipment;
The described module of cross exchange in length and breadth, be used to receive the original switching request that described up Switching Module sends, bit wide according to described original switching request, and the port number of the described module of cross exchange is in length and breadth duplicated described original switching request, make total bit wide of obtain after duplicating many parts exchange requests equal the output port number of the described module of cross exchange in length and breadth, whether the output port of checking each grouping of the described module of cross exchange in length and breadth links to each other with descending Switching Module, according to the result who checks, mask the exchange request of going to the output port that does not connect descending Switching Module, exchange request after the shielding processing is arbitrated, the arbitration result that obtains is exported to described up Switching Module, and transmit the cell that described up Switching Module is dispatched out according to described arbitration result.
A kind of packet switch apparatus, described device comprises:
The input request processing module, after being used to receive the original switching request of up Switching Module transmission, output port number according to the bit wide of described original switching request and this device is duplicated described original switching request, makes total bit wide of many parts of exchange requests that obtain after duplicating equal the output port number of this device;
Arbitration modules, be used to check whether the output port of each grouping links to each other with descending Switching Module, according to the result who checks, mask the exchange request of going to the output port that does not connect descending Switching Module, exchange request after the shielding is arbitrated, the arbitration result that obtains is exported to described up Switching Module;
Output allows processing module, is used for transmitting the cell of reception according to the result of described arbitration modules arbitration.
The beneficial effect of the technical scheme that the embodiment of the invention provides is:
By to original switching request duplicate the many parts of exchange requests that obtain, state according to output port carries out unitary arbitration to many parts of exchange requests, result according to arbitration dispatches cell, make between a plurality of groupings and realized intercommunication, expanded the port number of packet switching system, improved exchange capacity, simultaneously again can be compatible original SM system, reduced the cost of switching system, extensibility is strong.
Description of drawings
Fig. 1 is the structural representation of the big capacity fixed length grouping cell switching net that provides of prior art;
Fig. 2 is asymmetric three grades of switching network structural representations that the embodiment of the invention 1 provides;
Fig. 3 is the grouping exchange method flow chart that the embodiment of the invention 1 provides;
Fig. 4 is the grouping exchange method flow chart that the embodiment of the invention 2 provides;
Fig. 5 is the packet switching system schematic diagram that the embodiment of the invention 3 provides;
Fig. 6 is the packet switch apparatus schematic diagram that the embodiment of the invention 4 provides.
Embodiment
For the purpose, technical scheme and the advantage that make the embodiment of the invention is clearer, embodiment of the present invention is described further in detail below in conjunction with accompanying drawing.
The embodiment of the invention by to original switching request duplicate the many parts of exchange requests that obtain, state according to each grouping output port carries out unitary arbitration to many parts of exchange requests, result according to arbitration dispatches cell, make between a plurality of groupings and realized intercommunication, expand the port number of packet switching chip, and improved exchange capacity.
Present embodiment provides a kind of grouping exchange method, is used for cross exchange module in length and breadth, and to realize the exchange to the information in the shared buffer memory Switching Module, wherein, the shared buffer memory Switching Module comprises up Switching Module and descending Switching Module, and this method comprises:
Receive the original switching request that up Switching Module sends;
Request is duplicated to original switching, obtains many parts of exchange requests, and wherein, many parts of exchange requests are corresponding to a plurality of output ports of cross exchange module in length and breadth;
The state of the port that links to each other with descending Switching Module according to cross exchange module is in length and breadth arbitrated many parts of exchange requests, with the switching path of the cell determining to be required in the original switching request to exchange, export to up Switching Module after the arbitration result that obtains is converted to permission to the original switching request;
Receive the cell that up Switching Module sends; Result according to arbitration is transmitted to descending Switching Module with cell.
Wherein, the cross exchange module can be divided into a plurality of groupings according to the exchange needs in length and breadth, for example: with cross exchange module in length and breadth 16 input ports and 16 output ports being arranged is example, if exchange need be 4 * 4 exchange, be corresponding 4 output ports of 4 input ports of each grouping, then cross exchange module in length and breadth can be divided into 4 groupings; If exchange need be 8 * 8 exchange, then cross exchange module in length and breadth can be divided into 2 groupings, the function of each grouping is identical.
The asymmetric three grades of switching network structural representations that provide with Fig. 2 are that example describes below, and SM ply-yarn drill capacity is respectively 10G, 5G, 2.5G, 2.5G among Fig. 2.When capacity did not wait, the link number that the SM ply-yarn drill goes out did not wait, and 10G has gone out 4 links, and 2.5G has only gone out 1 link.The number of the SC grouping that connects of each SM ply-yarn drill does not just wait like this, if can not intercommunication between each grouping, will occur 10G SM can not with 5G+2.5G+2.5G SM intercommunication at full speed.
In embodiments of the present invention, SC comprises 4 groupings, it is one 4 * 4 exchange, be corresponding 4 output ports of 4 input ports of each grouping, the function of 4 groupings is identical, and 4 groupings of present embodiment are respectively the first grouping group0, the second grouping group1, the 3rd grouping group2 and the 4th grouping group3, and SC has 16 ports, can connect 16 SMi at most, SC physically is one 16 * 16 a arbitration modules; The exchange request of each SMi is 4bit (bit), corresponding 4 SMe.
Referring to Fig. 3, sending the exchange request with SMi0 to the first grouping group0 of SC is example, and this grouping exchange method specifically comprises:
S101:SMi0 sends the original switching request to SC;
SMi0 can send to SC with the original switching request by the arbitrary link in all links;
Wherein, the original switching request is 4 bits, and if 4 output ports in the corresponding grouping of each bit are bit[n]=1, the output port of then should exchange asking to be asked is output port n, expression need be to output port n transmission cell; For example, if the exchange request is: 0110 (corresponding successively output port 3,2,1,0), then expression need send cell to output port 2 and output port 1;
After S102:SC receives the original switching request, this original switching request is duplicated, obtain many parts of exchange requests according to the pattern that is provided with;
The pattern that is provided with in the present embodiment is that SC has 16 ports, the bit wide of each original switching request is 4 bits, so the exchange request after duplicating is 4 parts of exchange requests, promptly generate the exchange request of 16bit, make input port submit to ask to other grouping, for example, SMi0 is to 4 output ports of the corresponding group0 of 4bit request of the first grouping group0 proposition, if this 4bit request is not replicated, SMi0 can only select in 4 output ports of group0, after the 4bit request is replicated to the 16bit request, then can submit to ask to 4 groupings simultaneously, be that original 4bit request meeting is copied to the second grouping group1～the 4th grouping group3 simultaneously, 4 outbound ports to the second grouping group1～the 4th grouping group3 propose same request respectively, and the output port number that SMi0 can be selected has obtained expansion;
S103: according to the state of each grouping output port, SC carries out shielding processing to many parts of exchange requests that obtain after duplicating;
When SC had the part port not to be connected with SM, SC can only know the connection state of input port, can't learn the connection status of output port, and present embodiment can have multiple means to make SC obtain its output port state.For example: the mode that SC can be by auto-negotiation or the mode of software arrangements are determined the state of its output port, the mode of auto-negotiation refers to comprise link-state information (input link of SM is exactly the output link state of SC) or by manually by software interface configuration SC output port state in giving the cell of SC by SM, and the state of output port is kept on the SC, or be kept in the memory block that SC can read;
Can be learnt that by Fig. 2 the first grouping group0 has with 4 SMe and is connected, all input ports can be proposed 4bit exchange request to first group0 that divides into groups; The second grouping group1 only is connected with SMe1 with SMe0, will mask the exchange request of going to SMe2 and SMe3 so carry in the 4bit exchange request of group1; The 3rd grouping group2 only is connected with SMe0 with the 4th grouping group3, will mask the request of going to SMe1, SMe2 and SMe3 so carry in the exchange request of these two groupings;
S104: the arbitration modules unitary arbitration is submitted in the exchange request after the shielding processing;
Export to same output port simultaneously in order to prevent the cell that a plurality of inbound ports receive, and each as much as possible connecting, the SC of present embodiment carries out unitary arbitration by arbitration modules to the exchange request after shielding, wherein, arbitration modules can adopt arbitrary arbitration algorithm realization arbitration in the prior art, and present embodiment is not done concrete qualification to this.
The output port of S105:SC is handled arbitration result, converts arbitration result to the original switching request permission;
Because original exchange request has only 4bit, the scope of the output port of request is 0～3, and by above-mentioned duplicate exchange request after, the exchange request that obtains is 16bit, promptly the scope of the output port that can ask is 0～15, if directly above-mentioned arbitration result is returned to SMi, when the output port in the arbitration result greater than 3 the time, SMi can not discern, therefore need handle arbitration result, convert permission to, be about to output port in the arbitration result and be converted to numerical value in 0～3, specifically can realize by the following method greater than 3 numerical value to raw requests:
Output port in the arbitration result divided by the umber that duplicates back exchange request, is got remainder then, and this remainder is the output port after the conversion.
Because the exchange request after S102 duplicates in the present embodiment is 4 parts, also can be directly the high 2bit of the output port of arbitration result be abandoned, the output port that abandons behind high 2 bits is the permission output port that original 4bit is asked;
SC is in order to export to SMe from the output port (output port the arbitration result) of correspondence with the cell of receiving, SC will preserve the arbitration result that above-mentioned arbitration modules obtains;
S106:SC returns the allowance signal according to the arbitration result after changing to SMi0;
This allowance signal will return to SMi0 along the link that SMi0 sends above-mentioned exchange request;
S107:SMi0 after receiving and permitting signal comes out cell scheduling, by SC cell is sent to corresponding SMe.
Wherein, cell is a fixed length, and its length is to set according to the needs of actual conditions.
After SC receives the cell of SMi0 transmission, this cell is exported from the output port of correspondence according to the arbitration result of preserving.
The process that SC carries out shielding processing to many parts of exchange requests that obtain after duplicating among the above-mentioned S103 also can be handled by arbitration modules, arbitration modules is carried out shielding processing according to the state of each outbound port to the exchange request, identical among method and the above-mentioned S103, no longer describe in detail here.
In order to understand the processing procedure of said method better, the asymmetric three grades of switching network structural representations that provide with Fig. 2 are example below, suppose that the 4bit original switching request of each input port reception of SC is as follows:
Input port 0:0110
Input port 1:0101
Input port 2:1010
Input port 3:0011
Input port 4:1001
Input port 5:1101
Input port 8:1010
Input port 12:0101
Wherein, the 4bit in the exchange request of present embodiment is for the request of corresponding output port 3,2,1,0 successively, and according to different needs, the output port of the 4bit correspondence in the exchange request can be provided with arbitrarily, no longer describes in detail here;
Exchange request after each input port duplicates is as follows:
Input port 0:0110-0110-0110-0110
Input port 1:0101-0101-0101-0101
Input port 2:1010-1010-1010-1010
Input port 3:0011-0011-0011-0011
Input port 4:1001-1001-1001-1001
Input port 5:1101-1101-1101-1101
Input port 8:1010-1010-1010-1010
Input port 12:0101-0101-0101-0101
Wherein, be the request of corresponding output port 3,2,1,0 successively according to the 4bit in the exchange request, the 16bit of the exchange request after duplicating is corresponding output port 15,14 successively ...., 3,2,1,0 request; When the 4bit in the exchange request for output port when changing, the 16bit of the exchange request after duplicating also will correspondingly change, for example: the 4bit in the former exchange request is successively for output port 0,1,2,3 o'clock, then, the 16bit that exchange after duplicating is asked is correspondence 0 successively, 1,2 ..., 14,15.
Because the part output port does not connect SMe, for example output port 6,7,9,10,11,13,14,15, need the request of shielding to this output port, are about to the bit position 0 of corresponding these output ports in the request, and the result after the shielding is as follows:
Input port 0:0000-0000-0010-0110
Input port 1:0001-0001-0001-0101
Input port 2:0000-0000-0010-1010
Input port 3:0001-0001-0011-0011
Input port 4:0001-0001-0001-1001
Input port 5:0001-0001-0001-1101
Input port 8:0000-0000-0010-1010
Input port 12:0001-0001-0101-0101
The arbitration modules of SC is arbitrated the exchange request of each input port, supposes that the result of each output port of arbitration back is as follows:
0:5 represents that No. 0 input port of SC allows to establish a connection to output port No. 5;
1:2 represents that No. 1 input port of SC allows to establish a connection to output port No. 2;
2:1 represents that No. 2 input ports of SC allow to establish a connection to output port No. 1;
3:0 represents that No. 3 input ports of SC allow to establish a connection to output port No. 0;
4:8 represents that No. 4 input ports of SC allow to establish a connection to output port No. 8;
5:4 represents that No. 5 input ports of SC allow to establish a connection to output port No. 4;
8:3 represents that No. 8 input ports of SC allow to establish a connection to output port No. 3;
12:12 represents that No. 12 input ports of SC allow to establish a connection to output port No. 12;
The raw requests of each link input has only 4bit, and the scope of the output port of request is 0～3, directly above-mentioned arbitration result is returned to SMi, SMi can not discern, output port need be handled arbitration result, converts the permission to raw requests to, and the result after the processing is as follows:
Input port 0: output port 1
Input port 1: output port 2
Input port 2: output port 1
Input port 3: output port 0
Input port 4: output port 0
Input port 5: output port 0
Input port 8: output port 3
Input port 12: output port 0
Result after the processing represents that the input port on the left side can connect with the output port on the right;
Result after will changing by link corresponding then returns to corresponding SMi, carries out cell scheduling by SMi.
Present embodiment can make the exchange request in each grouping carry out having expanded the port number of SC alternately by exchange request is duplicated, and has improved exchange capacity, and the while again can compatible original SM chip, has reduced the cost of exchange chip, and extensibility is strong.
Present embodiment provides a kind of grouping exchange method, and present embodiment is an example with asymmetric three grades of switching network structural representations that Fig. 2 provides still, and wherein, the bandwidth of every link is 2.5G.Referring to Fig. 4, send the business of 10G below with the SMi0 desire, 5G is to SMe1, and 2.5G is to SMe2, and 2.5G as the shielding condition, simply describes the process of packet switching to SMe3:
S201:SMi0 respectively sends an exchange request to the first grouping group0 to the, four grouping group3 respectively, and request sends the business of 2.5G;
S202: after the first grouping group0 to the, four grouping group3 receive the exchange request, all the exchange request is duplicated, obtain 4 parts of exchange requests after duplicating;
S203: according to the shielding condition shielding processing is carried out in the exchange request that obtains after duplicating, the exchange request after the shielding processing sends to arbitration modules and arbitrates, and the result of present embodiment arbitration is:
Allow SMi0 to be connected with SMe1 with the second grouping group1 by the first grouping group0;
Allow SMi0 to be connected with SMe3 with SMe2 by the first grouping group0;
Wherein, shielding processing can be carried out according to the mode that embodiment 1 provides;
The output port of S204:SC is handled arbitration result, converts arbitration result to raw requests permission;
Arbitration result after present embodiment is handled is:
Allow SMi0 to be connected with SMe1 by the first grouping group0;
Allow SMi0 to be connected with SMe1 by the second grouping group1;
Allow SMi0 to be connected with SMe2 by the 3rd grouping group2;
Allow SMi0 to be connected with SMe3 by the 4th grouping group3;
S205:SC returns the allowance signal according to the result of arbitration to SMi0;
S206:SMi0 will send the business of 2.5G respectively to SC from four output ports after receiving and permitting signal;
After S207:SC receives the 10G business that SMi0 sends, on the first grouping group0 and the second grouping group1, SMe1 is proposed the transmission request, on the first grouping group0, SMe2 and SMe3 are proposed the transmission request according to the arbitration result of arbitration modules;
S208:SC sends to SMe1 by link corresponding with the 5G business after receiving the affirmation response of SMe1, SMe2 and SMe3, and the 2.5G business sends to SMe2, and the 2.5G business sends to SMe3.
The said method of present embodiment can arbitrate the exchange request by duplicating the exchange request that SMi0 sends in four groupings of SC, and then realizes that big capacity SMi sends to the linear speed of low capacity SMe.
The method that provides according to present embodiment, if it is professional to SMe0 that the SMi1 desire sends 5G, when SMi2 and SMi3 send the 2.5G business to SMe0, SMi1 sends to the exchange request that the exchange request of group0 and group1 and SMi2 and SMi3 send to group0 and will expand in 4 groupings by copy operation, in 4 groupings, all the transmission request is proposed SMe0, and then the business that SMi1, SMi2 and SMi3 desire are sent sends on the SMe0, realizes that low capacity SMi sends to the linear speed of big capacity SMe.Simultaneously, present embodiment can be supported multiple application model setting, promptly support the various connection modes of SMi or SMe and SC, the SM core number software arrangements that support to connect, and the mode of mode that SC can be by auto-negotiation or software arrangements is determined the state of its output port.
Referring to Fig. 5, the embodiment of the invention also provides a kind of packet switching system, this system comprises: shared buffer memory Switching Module 301 and cross exchange module 302 in length and breadth, wherein, shared buffer memory Switching Module 301 and in length and breadth cross exchange module 302 can be integrated in one or several chip;
Wherein, shared buffer memory Switching Module 301 comprises:
Up Switching Module 301a is used for to cross exchange module 302 transmission original switching requests in length and breadth, and after receiving the allowance signal of original switching request, dispatches the concurrent unit of delivering letters and give cross exchange module 302 in length and breadth;
Descending Switching Module 301b is used to receive the cell of cross exchange module 302 forwardings in length and breadth, and cell is sent to corresponding equipment.
The cross exchange module 302 in length and breadth, be used to receive the original switching request that up Switching Module 301a sends, request is duplicated to original switching, obtains many parts of exchange requests, wherein, many parts of exchange requests are corresponding to a plurality of output ports of cross exchange module 302 in length and breadth; State according to the port that links to each other with descending Switching Module 301b is arbitrated many parts of exchange requests, the arbitration result that obtains is exported to up Switching Module 301a, and transmit the cell that up Switching Module 301a dispatches out according to arbitration result.
During system that the specific implementation present embodiment provides, can adopt SMi, SC and SMe among Fig. 2 that embodiment 1 provides to realize, no longer describe in detail here.
The system that present embodiment provides by to original switching request duplicate, obtain many parts of exchange requests, the bit wide of exchange request is equated with the number of output port, please seek unification to many parts of exchanges then and carry out arbitration process, the group of striding when having realized packet switching exchanges, and has expanded the port number of cross exchange module in length and breadth, improved exchange capacity, simultaneously again can compatible original SM ply-yarn drill, reduced the cost of switching system, extensibility is strong.
Embodiment 4
Referring to Fig. 6, present embodiment also provides a kind of packet switch apparatus, and this device comprises:
Input request processing module 401, after being used to receive the original switching request of up Switching Module transmission, request is duplicated to original switching, obtains many parts of exchange requests, wherein, the many parts of a plurality of output ports that exchange request corresponding to this device;
Output allows processing module 403, is used for transmitting the cell of reception according to the result of arbitration modules 402 arbitrations.
After input request processing module 401 specifically is used to receive the original switching request of up Switching Module transmission, according to the output port number of the bit wide of original switching request and this device to original switching request duplicate, make total bit wide of many parts of exchange requests that obtain after duplicating equal the output port number of this device.
Packet switching system described in the embodiment of the invention can be integrated in one or several chip.
Further, arbitration modules 402 comprises:
Arbitration unit 402c is used for the exchange request after the screen unit 402b shielding is arbitrated, and obtains arbitration result;
Converting unit 402d is used for arbitration result with arbitration unit 402c and is converted to permission to raw requests;
Permit signal transmitting unit 402e, be used for returning the allowance signal to up Switching Module according to the arbitration result after the converting unit 402d conversion.
The packet switching system that present embodiment provides can possess the function of SC among the foregoing description 1 or the embodiment 2, no longer describes in detail here.
Present embodiment by to original switching request duplicate, the bit wide of exchange request is equated with the number of output port, please seek unification to many parts of exchanges then and carry out arbitration process, realized not increasing under SM and the situation that the SC physical link is connected the linear speed intercommunication of asymmetric traffic in the single SC system.
Above embodiment solved in the prior art between the grouping among the SC can not intercommunication problem.For example, connectivity port number between big capacity SM and the SC is more than the connectivity port number of low capacity SM, the grouping number of therefore big capacity SM connection SC can be more than the grouping number of low capacity SM connection SC, exchange and can not stride group between each grouping in the prior art, making between 2 SM that do not connect same grouping can not intercommunication, can address this problem by the block exchange mode in the above embodiment of the present invention.
All or part of content in the technical scheme that above embodiment provides can realize that its software program is stored in the storage medium that can read by software programming, storage medium for example: the hard disk in the computer, CD or floppy disk.
The above only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (5)
1. a grouping exchange method is used for cross exchange module in length and breadth, and to realize the exchange to the information in the shared buffer memory Switching Module, described shared buffer memory Switching Module comprises up Switching Module and descending Switching Module, it is characterized in that, described method comprises:
Receive the original switching request that described up Switching Module sends;
Bit wide according to described original switching request, and the port number of the described module of cross exchange in length and breadth duplicates described original switching request, makes total bit wide of many parts of exchange requests that obtain after duplicating equal the output port number of the described module of cross exchange in length and breadth;
Whether the output port of checking each grouping of the described module of cross exchange in length and breadth links to each other with descending Switching Module;
Result according to checking masks the exchange request of going to the output port that does not connect descending Switching Module;
Exchange request after the shielding processing is arbitrated,, the arbitration result that obtains is exported to described up Switching Module to select a switching path in the described many parts of exchange requests as cell;
Receive the cell that described up Switching Module sends;
Result according to described arbitration is transmitted to described descending Switching Module with described cell.
2. grouping exchange method as claimed in claim 1 is characterized in that, describedly the arbitration result that obtains is exported to described up Switching Module comprises:
Described arbitration result is converted to permission to described original switching request;
Return the allowance signal according to the arbitration result after the conversion to described up Switching Module.
3. a packet switching system is characterized in that, described system comprises: shared buffer memory Switching Module and cross exchange module in length and breadth;
Described shared buffer memory Switching Module comprises:
Up Switching Module is used for sending the original switching request to the described module of cross exchange in length and breadth, and after receiving the allowance signal of described original switching request, dispatches the concurrent unit of delivering letters and give the described module of cross exchange in length and breadth;
Descending Switching Module is used to receive the cell of the described module forwards of cross exchange in length and breadth, and described cell is sent to corresponding equipment;
The described module of cross exchange in length and breadth, be used to receive the original switching request that described up Switching Module sends, bit wide according to described original switching request, and the port number of the described module of cross exchange is in length and breadth duplicated described original switching request, make total bit wide of obtain after duplicating many parts exchange requests equal the output port number of the described module of cross exchange in length and breadth, whether the output port of checking each grouping of the described module of cross exchange in length and breadth links to each other with descending Switching Module, according to the result who checks, mask the exchange request of going to the output port that does not connect descending Switching Module, exchange request after the shielding processing is arbitrated, the arbitration result that obtains is exported to described up Switching Module, and transmit the cell that described up Switching Module is dispatched out according to described arbitration result.
4. a packet switch apparatus is characterized in that, described device comprises:
The input request processing module, after being used to receive the original switching request of up Switching Module transmission, output port number according to the bit wide of described original switching request and this device is duplicated described original switching request, makes total bit wide of many parts of exchange requests that obtain after duplicating equal the output port number of this device;
Arbitration modules, be used to check whether the output port of each grouping links to each other with descending Switching Module, according to the result who checks, mask the exchange request of going to the output port that does not connect descending Switching Module, exchange request after the shielding is arbitrated, the arbitration result that obtains is exported to described up Switching Module;
Output allows processing module, is used for transmitting the cell of reception according to the result of described arbitration modules arbitration.
5. packet switch apparatus as claimed in claim 4 is characterized in that, described arbitration modules comprises:
Inspection unit is used to check whether the output port of each grouping links to each other with descending Switching Module;
Screen unit is used for the result according to described inspection unit inspection, masks the exchange request of going to the output port that does not connect descending Switching Module;
Arbitration unit is used for the exchange request after the described screen unit shielding is arbitrated, and obtains arbitration result;
Converting unit is used for arbitration result with described arbitration unit and is converted to permission to described raw requests;
Permit signal transmitting unit, be used for returning the allowance signal to up Switching Module according to the arbitration result after the described converting unit conversion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008101166314A CN101631067B (en) | 2008-07-14 | 2008-07-14 | Method, device and system for packet switching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008101166314A CN101631067B (en) | 2008-07-14 | 2008-07-14 | Method, device and system for packet switching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101631067A CN101631067A (en) | 2010-01-20 |
| CN101631067B true CN101631067B (en) | 2011-12-21 |
Family
ID=41576021
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2008101166314A Active CN101631067B (en) | 2008-07-14 | 2008-07-14 | Method, device and system for packet switching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101631067B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004112743A (en) * | 2002-07-24 | 2004-04-08 | Kddi Corp | Route-selecting method |
| CN1783828A (en) * | 2004-12-02 | 2006-06-07 | 华为技术有限公司 | Space division exchanging device and information element exchanging method |
-
2008
- 2008-07-14 CN CN2008101166314A patent/CN101631067B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004112743A (en) * | 2002-07-24 | 2004-04-08 | Kddi Corp | Route-selecting method |
| CN1783828A (en) * | 2004-12-02 | 2006-06-07 | 华为技术有限公司 | Space division exchanging device and information element exchanging method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101631067A (en) | 2010-01-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9426099B2 (en) | Router, method for controlling router, and program | |
| CN101873253B (en) | Buffered crossbar switch system | |
| CN101841420B (en) | Network-on-chip oriented low delay router structure | |
| AU748133B2 (en) | Multi-port internally cached drams | |
| CN100531085C (en) | Method for accessing into control address table spacing using Ethernet switch medium | |
| CN102868604B (en) | Two-dimension Mesh double buffering fault-tolerant route unit applied to network on chip | |
| CN101414973B (en) | Method and apparatus for transmitting data packet | |
| CN110995598B (en) | Variable-length message data processing method and scheduling device | |
| CN102063408B (en) | Data bus in multi-kernel processor chip | |
| CN103136141A (en) | High speed interconnection method among multi-controllers | |
| CN104461966B (en) | A kind of data buffer storage transmission method and its controller based on Nand Flash chips | |
| CN111522775A (en) | Network-on-chip routing device and control method thereof | |
| CN1656470A (en) | Inter-chip processor control plane communication | |
| TW201138379A (en) | Directly providing data messages to a protocol layer | |
| CN103106177B (en) | Interconnect architecture and method thereof on the sheet of multi-core network processor | |
| US11436185B2 (en) | System and method for transaction broadcast in a network on chip | |
| CN101631067B (en) | Method, device and system for packet switching | |
| US20040078459A1 (en) | Switch operation scheduling mechanism with concurrent connection and queue scheduling | |
| CN106547707A (en) | Cluster memory storage concurrent access Local Priority switched circuit in AP | |
| EP0322116B1 (en) | Interconnect system for multiprocessor structure | |
| US20020146034A1 (en) | Self-route multi-memory expandable packet switch with overflow processing means | |
| US7254139B2 (en) | Data transmission system with multi-memory packet switch | |
| US7272151B2 (en) | Centralized switching fabric scheduler supporting simultaneous updates | |
| CN101488119B (en) | Address interpretation method, apparatus and single-board | |
| US7236497B2 (en) | Facilitating arbitration via information associated with groups of requesters |
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 |