WO2010081365A1

WO2010081365A1 - Processing method and system for preventing congestion

Info

Publication number: WO2010081365A1
Application number: PCT/CN2009/075635
Authority: WO
Inventors: 赖伟
Original assignee: 中兴通讯股份有限公司
Priority date: 2009-01-16
Filing date: 2009-12-16
Publication date: 2010-07-22
Also published as: CN101783763B; CN101783763A

Abstract

A processing method and system for preventing congestion are provided, wherein the processing method for preventing congestion includes: the input ends send the size information of each queue to be sent to the corresponding output ends; the input ends receive and save the queue information from each output end separately, wherein the queue information includes the port information of the output ends and the size information of the queues, and the size information of the queues indicates the sum of size of each queue, and the input ends process the received messages to be in the queue based on the queue information and preset rules. By the present invention, the message congestion circumstance of the output ends is relieved in advance, and thus the problem that messages are processed incorrectly in output ends is avoided.

Description

TECHNICAL FIELD The present invention relates to the field of communications, and in particular to a method and system for preventing congestion. BACKGROUND OF THE INVENTION In essence, an Internet network is a collection of heterogeneous networks connected through a core router. As network services develop, the capacity of core routers will further expand. FIG. 1 is a schematic diagram of the architecture of a current core router. The core switching module (including the switching access chip and the switching chip in FIG. 1) is a bridge connecting an input end (ingress) and an output end (Egress), and is a router implementation grouping. Forwarded core device. Typically, weighted random early discards are applied at each input of the core switching network (Weighted

The Randomly Early Discard (WRED) detection function is used to avoid congestion on the input side. The WRED detection processes the packets received by the input according to the queue parameters, weights, or cache occupancy, including discarding or adding to the corresponding queue. in. The WRED detection can adjust the sending traffic of the input end, so that the entire network can be prevented from entering the congestion state and all the packets are discarded. At present, the WRED detection implementation algorithm is relatively mature. However, since the input-based WRED detection only acts on the input end and cannot be applied to the output end, there may still be packet congestion at the output end. Once packet congestion occurs, the packet may need to be discarded. For example, FIG. 2 shows the case where congestion occurs at the output end in the core switching network. As shown in FIG. 2, two data streams of the same level (represented by Flow A and Flow B, respectively) are from device A and device B, respectively. The size is 1G, and the two data streams are simultaneously output to one port (that is, the switching access chip shown on the right side of FIG. 2), and the output port has a traffic of 1G, assuming that the data packet in Flow A is a committed bandwidth. The data packet in Flow B is the excess bandwidth. That is, the priority of the data packet in Flow A is greater than the priority of the data packet in Flow B. According to the current processing, half of the data packets in Flow A will be discarded. High priority data will be discarded, which is undoubtedly detrimental to the operation and performance of the system. SUMMARY OF THE INVENTION The present invention has been made in view of the problem that an output terminal cannot properly process a message due to packet congestion occurring at the above output. To this end, the present invention aims to provide an improved anti-congestion scheme to solve the above problem. In order to achieve the above object, according to an aspect of the present invention, an anti-congestion processing method is provided, which is applied to a system layer, wherein the system layer includes an input end and an output end. The anti-congestion processing method according to the present invention includes: the input end sends the size information of each queue to be sent to the corresponding output end; the input end respectively receives and saves the queue information from each output end, wherein the queue information includes the port of the output end Information, queue size information, and queue size information indicate the sum of the queue sizes at the output; according to the queue information and the predetermined rule, the input end processes the received message to be queued. The predetermined rule is: performing weighted random early discard detection on the waiting queue of the received message. The operation of sending the size information of each queue to be sent to the corresponding output end includes the following: the input end periodically sends the size information of each queue to be sent to the corresponding output end; the input end occurs in the size of the queue. When the change is made, the size information of each queue to be sent is sent to the corresponding output. And after the size information of each queue to be sent is sent to the corresponding output, each output periodically or aperiodically feeds back the queue information to the input end. In addition, if the queue size indicated by the latest feedback queue information of the output terminal is larger than the queue size indicated by the previously saved queue information, the previously saved queue information is updated using the latest feedback queue information. Specifically, according to the queue information and the predetermined rule, the processing performed by the input end on the received packet to be queued includes: determining, according to the destination address of the packet to be queued, the queue that the packet needs to be enqueued, and further determining the need The output corresponding to the queue of the enqueue; obtains the queue information of the corresponding output; discards or enqueues the packet according to the queue size indicated by the queue information and the result of the weighted random early discard detection. In order to achieve the above object, according to another aspect of the present invention, an anti-congestion system is provided, the anti-congestion system including an input end and an output end, wherein the input end includes a first for receiving a message Receive module. According to the anti-congestion system of the present invention, the input terminal further includes: a first sending module, configured to send size information of each queue to be sent to a corresponding output end; and a second receiving module, configured to separately receive and save each of the The queue information of the output end, wherein the queue information includes the port information of the output end and the queue size information, and the queue size information indicates the sum of the size of each queue of the output end; the processing module is configured to use the queue information received by the second receiving module according to the predetermined rule. Processing the packet to be queued received by the first receiving module; the output terminal includes: a third receiving module, configured to receive information sent by the first sending module; and an operation module, configured to receive according to the third receiving module The information is used to calculate the queue size information of the output end; the second sending module is configured to send the queue size information of the output end calculated by the computing module to the second receiving module. Preferably, the input terminal further includes: an update module, configured to update the previously saved queue information. The first timer is configured to periodically send size information of each queue to be sent to the output end. And the output terminal further includes: a second timer, configured to periodically feed back queue information to the input end. By means of at least one of the foregoing technical solutions, the invention interacts with the input end, so that the received message is processed at the input end, which can relieve the congestion of the output end of the message, thereby avoiding the error of the output end. Handling messages. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are set to illustrate,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, In the drawings: FIG. 1 is a schematic structural diagram of a core router in the related art; FIG. 2 is a schematic diagram of a congestion state of an output port in a core switching network in the related art; FIG. 3 is an implementation according to an embodiment of the present invention. FIG. 4 is a block diagram of an anti-congestion system according to an embodiment of the present invention; and FIG. 5 is a schematic structural diagram of an anti-congestion system according to an embodiment of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described in detail with reference to the accompanying drawings in conjunction with the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments may be combined with each other if they do not conflict. Method Embodiment In the embodiment of the present invention, an anti-congestion processing method is provided. The method is applied to a system layer, where the system layer herein may be the core switching network, including an input end and an output end. FIG. 3 is a flowchart of an anti-congestion processing method according to an embodiment of the present invention. As shown in FIG. 3, the anti-congestion processing method includes the following steps (steps S302-S306). Step S302, the input end sends the size information of each queue to be sent to the corresponding one or more output ends. In the specific implementation process, the input end can send the size information to the following two types: (method 1) The size information of each queue to be sent can be periodically sent to the corresponding output end, that is, at the input end. A timer is set to send the size information of each queue to be sent to the corresponding output when the timer expires. (Method 2) When the size of the queue changes, the size information of each queue to be sent is sent to The corresponding output. Step S304, the input end respectively receives and saves the queue information from each output end, wherein the queue information may include port information of the output end and queue size information, and the queue size information indicates a sum of the size of each queue of the output end. In the step S304, after receiving the information sent by the input end in the foregoing step S302, each output first calculates the queue size information (ie, the above-mentioned queue information) that needs to be processed on the local side, and then periodically or The queue information is fed back to each input aperiodically, so that the input can know in advance the data traffic that needs to be processed at the output. In the specific implementation process, the input port may be configured with a destination port Queue Size Table, where the address of each output terminal and the queue information fed back by each output end are used to maintain the traffic volume of each output end. The input side can find the queue information fed back by the output corresponding to the address according to the address in the table. The initial value of the queue information at each output in the table can be set to 0 to update according to the queue information of the received output. If the queue size indicated by the latest feedback queue information is larger than the queue size indicated by the previously saved queue information, the latest feedback queue information is used to the previously saved queue information. The update is performed, that is, each time the input end receives the queue information (which may be in the form of a cell), the queue size indicated by the received queue information is compared with the value of the corresponding address in the table, if the comparison result is greater than The value in the corresponding address is updated with the received queue information, otherwise the table is not updated. Step S306, according to the queue information and the predetermined rule, the input end processes the received packet to be queued. The predetermined rule here can preferably be implemented by the weighted random early discard detection technology in the prior art, which technology is relatively mature at present, and will not be mentioned here. Specifically, the input end receives a message, first determines the destination address of the packet to be queued, and further determines the output end of the queue that needs to be enqueued, thereby obtaining the Queue information of the corresponding output; according to the queue size indicated by the queue information, combined with the weighted random early discard detection technology, discarding or enqueuing the essay. In this way, the received message is processed in advance at the input end, and the congestion of the message at the output end can be alleviated in advance, thereby avoiding the problem of processing the message incorrectly at the output end. It can be seen from the above description that the size information of each queue to be sent is sent to the corresponding output end at the input end, so that the output end can feed back the size of the message to be processed on the side, so that the input end can be based on the output end. The feedback situation, combined with the weighted random early discard detection technology, processes the received 4 texts (including discarding or entering the corresponding queue;). The invention will now be further described from the perspective of a system layer, i.e., embodiments of the invention are described from the input side and the output side, respectively. (1) Input side step 1 , the input side sends the size of each queue (in the form of a cell) to each output through the core switching network to notify the output of the size of the traffic to be processed; this step corresponds to the above step S302 Step 2, the input side sets the destination port queue size table, and sets the value in the table according to the received "port queue size cell". The input side compares the queue size indicated in the received port queue size cell with the value in the corresponding address. If the result of the comparison is greater than the value in the corresponding address, the queue size is indicated in the corresponding address. The value is updated, otherwise, the value in the corresponding address is not updated; the step 4 is corresponding to the above step 4 S304; Step 3, the input side processes the packet to be enqueued, and uses the destination address carried by the packet The queue size of the corresponding output terminal in the destination port queue size table is searched, and the packet is processed in conjunction with the weighted random early discarding technique to avoid congestion at the output end; this step corresponds to the above step S306.

(2) Output side Step 1 , The output terminal can set an output queue size table locally ( Output Port

Queue Size Table ). Since an output can have multiple queues, these queues correspond to different inputs. Therefore, one output may receive multiple "queue size cells", and the output needs to have multiple "queue size cells". The indicated queue size is added to get the total queue size, and this value is filled into the output queue size table; Step 2, the value in the output queue size table is carried on the "port queue size cell", through the core The switching network is sent to each input terminal to notify each input end of the local (ie, output) message to be processed; Step 3, the output end optimizes the output queue size table, that is, periodic or The output queue size table is set to 0 aperiodically. Similar to step 2 of the above input, the output compares the calculated total queue size with the value in the output queue size table each time. If the result of the comparison is greater than the value in the table, the total queue size is calculated. Update, otherwise, the values in the table are not updated. It should be noted that, for convenience of description, in the above, the technical solutions of the method embodiments of the present invention are shown and described in the form of steps, and the steps shown above may be in a computer system such as a set of computer executable instructions. Executed in. Although the logical order is shown in the above description, in some cases, the steps shown or described may be performed in a different order than the ones described herein. Apparatus Embodiments In an embodiment of the present invention, an anti-congestion system is provided, which may be preferably used to implement the method provided by the foregoing method embodiments. 4 is a block diagram of an anti-congestion system according to the present embodiment. As shown in FIG. 4, the anti-congestion system includes an input terminal 1 and an output terminal 2. In practice, one input terminal can correspond to multiple output terminals, and one output The terminal can correspond to a plurality of input terminals. For convenience of description, an output terminal and an input terminal are illustrated in FIG. 4 as an example. As shown in FIG. 4, the input terminal 1 includes: a first receiving module 10, a first sending module 12, a second receiving module 14, and a processing module 16. The output terminal 2 includes: a third receiving module 20, an arithmetic module 22 and the second transmitting module 24, each of which is described below. The first receiving module 10 is configured to receive the size information of each queue to be sent to the corresponding output end, and the second receiving module 14 is configured to separately receive and save the data from each The queue information of the output end, wherein the queue information includes the port information of the output end, the queue size information, and the queue size information indicates the sum of the queue sizes of the output end; the processing module 16 is connected to the first receiving module 10 and the second receiving module 14, For processing the packet to be queued received by the first receiving module 10 according to the predetermined rule and the queue information received by the second receiving module 14, the third receiving module 20 is configured to receive the sending from the first sending module 12 The operation module 22 is connected to the third receiving module 20 for calculating the queue size information of the output terminal according to the information received by the third receiving module 20; the second sending module 24 is connected to the computing module 22 for using the computing module 22 The calculated queue size information of the output is sent to the second receiving module 14. The predetermined rule here may be the weighted random early discard detection technique in the above method embodiment. In the specific implementation process, the input terminal 1 may further include an update module (not shown) for updating the previously saved queue information, and each of the output terminal 1 and the input terminal 2 may include a timer. (not shown in the figure), wherein the timer of the input terminal 1 is used to periodically send the size information of each queue to be sent to the output end, and the timer of the output end 2 is used to periodically The queue information is fed back to the input. It can be seen from the above description that the first sending module 12 and the second sending module 24 enable the output end and the input end to interact, and the input end can perform corresponding processing on the received 4 艮 text, which can be relieved in advance. The congestion of the packets at the output end can avoid the problem of the packets being processed incorrectly at the output end. Similar to the above-described method embodiments, in the present embodiment, the present invention is further described from the perspective of the system layer, that is, the embodiments of the present invention are described from the input side and the output side, respectively. FIG. 5 is a schematic structural diagram of an anti-congestion system according to an embodiment of the present invention. As shown in FIG. 5, the structures of the input side and the output side are described in detail below in conjunction with actual applications.

(1) The input side virtual destination port queue module 501, corresponding to the second receiving module 14, is configured to save the destination port queue size table, so that the packet to be queued can be in virtual destination according to the destination address carried by the packet. The port queue module 301 finds the corresponding destination port queue size; the weighted random early discarding module 502 corresponds to the processing module 16 described above, and is configured according to the foregoing report. The queue number of the inbound queue and the corresponding destination port queue size are selected as the discarding policy, and the received packet is processed as a result of the discarding policy. If the result of the discarding policy is discarded, the packet is discarded; the virtual queue module 503 And sending, by the sending module 504, the sending module 504 of the input side is responsible for sending the packet and the queue size information (in the format of the cell) respectively. Switch to the switch chip. The virtual queue module 503 and the sending module 504 correspond to the first sending module 12 described above.

(2) an output side cell classification module 505, configured to perform classification on various cells received from the switch chip; a virtual destination port queue module 506 corresponding to the third receiving module 20 and the operation module 22, The received queue size information is updated according to the output end corresponding to the queue size information, and the output queue size table is updated, and the value in the output queue size table may be sent in a cell format periodically or non-periodically. To the switch chip, the port virtual queue module 507 is configured to enqueue the message sent by the input end, and wait for the output; the port module 508 is configured to schedule the output message to complete the message exchange process. In summary, according to the foregoing embodiment of the present invention, the input end sends the size information of each queue to be sent to the corresponding output end, and then the output end returns the queue information to the output end, so that the output end can be based on the queue information. The received message is processed. Compared with the prior art, the interaction between the input end and the output end can relieve the congestion of the output end in advance at the input end, thereby preventing the output end from being erroneously processing the message. Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general-purpose computing device, which can be concentrated on a single computing device or distributed over a network composed of multiple computing devices. Alternatively, they may be implemented by program code executable by the computing device, such that they may be stored in the storage device by the computing device, or they may be separately fabricated into individual integrated circuit modules, or they may be Multiple modules or steps are made into a single integrated circuit module. Thus, the invention is not limited to any specific combination of hardware and software. The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the scope of the present invention are intended to be included within the scope of the present invention.

Claims

Claim

An anti-congestion processing method is applied to a system layer, where the system layer includes an input end and an output end, and the method includes:

The input end sends the size information of each queue to be sent to the corresponding output end; the input end receives and saves the queue information from each output end, wherein the queue information includes the port information of the output end and the queue size. Information, the queue size information indicating a sum of sizes of the queues at the output end;

According to the queue information and the predetermined rule, the input terminal processes the received message to be queued.

2. The method according to claim 1, wherein the input end sends the size information of each queue to be sent to the corresponding output end, including one of the following:

The input end periodically sends size information of each queue to be sent to the corresponding output end;

The input end sends the size information of each queue to be sent to the corresponding output end when the size of the queue changes.

The method of claim 2, wherein after the inputting, by the input end, the size information of each queue to be sent to the corresponding output end, the method further includes: The output periodically or non-periodically feeds back the queue information to the input.

4. The method according to claim 3, further comprising:

If the queue size indicated by the most recently fed back queue information is larger than the queue size indicated by the previously saved queue information, the previously saved queue information is updated using the latest feedback queue information.

The method according to claim 1, wherein the predetermined rule is: performing weighted random early discard detection on a queue to be received of the received message. The method according to claim 5, characterized in that, according to the queue information and the predetermined rule, the inputting, by the input terminal, the received message to be queued includes:

Determining, according to the destination address of the packet to be enqueued, the queue that the packet needs to be enqueued, and further determining the output end corresponding to the queue that needs to be enqueued;

Obtaining queue information of the corresponding output end;

The packet is discarded or enqueued according to the queue size indicated by the queue information and the result of the weighted random early discard detection. An anti-congestion system, the anti-congestion system comprising an input end and an output end, the input end comprising a first receiving module for receiving a message, wherein

The input terminal further includes:

a first sending module, configured to send size information of each queue to be sent to a corresponding output end;

a second receiving module, configured to receive and save the queue information from each output end, where the queue information includes port information of the output end and queue size information, where the queue size information indicates a sum of the size of each queue of the output end ;

a processing module, configured to process, according to a predetermined rule and the queue information received by the second receiving module, a packet to be queued received by the first receiving module; the output end includes:

a third receiving module, configured to receive information sent by the first sending module, and an operation module, configured to calculate queue size information of the output end according to information received by the third receiving module;

And a second sending module, configured to send the queue size information of the output end calculated by the operation module to the second receiving module. The system of claim 7, wherein the input terminal further comprises: an update module, configured to update the previously saved queue information. The system according to claim 7, wherein the input terminal further comprises: a first timer, configured to periodically send size information of each queue to be sent to the output end.

The system according to claim 7, wherein the output terminal further comprises: a second timer, configured to periodically feed back the queue information to the input end.