WO2018166482A1 - Congestion management method and device for switched network, and computer storage medium - Google Patents

Abstract

Disclosed is a congestion management method for a switched network, comprising: reading a switching state table; selecting all first-order switching devices in a rack J according to the switching state table, and obtaining K links corresponding to the devices; selecting L links from the K links, gradually accumulating bandwidths of the L links, and using two adjacent accumulation results as a first bandwidth accumulation value and a second bandwidth accumulation value, the first bandwidth accumulation value being a bandwidth accumulation value of a links, and the second bandwidth accumulation value being a bandwidth accumulation value of b links; and closing links in the K links except for the a links when it is determined that the first bandwidth accumulation value is less than a preset bandwidth, and the second bandwidth accumulation value is greater than or equal to the preset bandwidth. Also disclosed are a congestion management device and a computer storage medium.

Description

Congestion management method and device for switching network, computer storage medium

Cross-reference to related applications

The present application is based on a Chinese patent application filed on Jan. 14, 2017, the entire disclosure of which is hereby incorporated by reference.

Technical field

The present invention relates to congestion management technologies for switching networks, and in particular, to a congestion management method and apparatus for a switching network, and a computer storage medium.

Background technique

The switching system is a core component of the packet switching device, and is composed of a switching access part and a switching network. The switching access part includes a source switching access unit (SAU) and a destination SAU, and the switching network is composed of multiple Switching Units (SUs) are connected to each other. The source SAU is interconnected with the SU through a high-speed link, and the SUs are interconnected by a high-speed link, and the SU is interconnected with the destination SAU through a high-speed link.

In a switched network, each SU is connected to each other, and according to its connection mode, it can be divided into a single-level networking structure and a three-level non-blocking (CLOS) networking structure. The third-level CLOS network is shown in Figure 1. It consists of three-level SUs. There are service chassis 1, service chassis 2, and central chassis 65. The packets from the source switching access device on service chassis 1 are sent. The element passes through the SU on the service chassis 1 to reach the central chassis 65, and the SU through the central chassis 65 exchanges the SU to the SU on the service chassis 2, and then exchanges with the SU on the service chassis 2 to finally reach the service chassis. 2 SAU for each purpose. For the sake of distinction, the SU connected to the source SAU on the service chassis is called the first-level SU, the SU on the central chassis is called the second-level SU, and the SU connected to the destination SAU on the service chassis is called the third-level SU.

An asymmetric switching network means that there is a bandwidth inconsistency between two levels adjacent to the switching network, that is, the bandwidth of all the nodes in any one level is inconsistent. The third-level CLOS asymmetric switching network can be divided into asymmetry between the source switching access device on the service chassis and the first-level switching device, and the first-level switching device and the second-level switching device are asymmetric, and the second level The switching device is asymmetric with the third-level switching device and the third-level switching device and the destination switching access device on the service chassis are asymmetric; here, one service chassis can place multiple switching devices, and each switching device A plurality of SUs may be included, and therefore, the first stage SU described above is also referred to as a first stage switching device SU. As shown in FIG. 2, the input and output bandwidths of the first-stage SU and the second-stage SU on the service chassis and the central chassis are the same, but at the third-level SU2# on the service chassis 3, The input bandwidth is greater than the output bandwidth, where the dotted arrow line indicates that the bandwidth is low or the link is disconnected, and in the case where the input traffic is sufficient, local congestion occurs in the third-level SU2# on the service chassis 3. When one or more SUs in the switching network are partially congested, the cell switching speed will lag behind other SUs, which will eventually cause the entire network traffic to drop. When the congestion is severe, packet loss may occur, and system performance may be degraded.

Summary of the invention

The embodiment of the present invention is to provide a congestion management method and device for a switching network, which can effectively solve the congestion or packet loss problem of the switching network.

An embodiment of the present invention provides a congestion management method for a switching network, where the method includes: reading a switching state table, where the switching state table is an M row and an N column state table; wherein the switching state table is 1 to The row order of the M rows represents the number of the different first-level switching devices, and the data of the first to Nth columns of the switching state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the representation and Link information in the congestion management apparatus connected to the first-stage switching device, where M and N are positive integers;

Selecting, according to the exchange state table, all L first-level switching devices belonging to the rack J, and acquiring K links corresponding to the L first-level switching devices; wherein, L and J are a positive integer less than M, the K being a positive integer less than the product of L and N;

Selecting L links from the K links, and accumulating the bandwidths of the L links one by one, and using the two adjacent accumulated results as the first bandwidth accumulated value and the second bandwidth accumulated value respectively; The first bandwidth accumulation value is a bandwidth accumulation value of a link, and the second bandwidth accumulation value is a bandwidth accumulation value of the b links, where the a link and the b link are both a link in the L links, wherein a and b are both positive integers less than or equal to L, and b=a+1;

When it is determined that the first bandwidth accumulated value is less than a preset bandwidth, and the second bandwidth accumulated value is greater than or equal to the preset bandwidth, closing a chain other than the a links in the K links road.

An embodiment of the present invention further provides a congestion management apparatus for a switching network, where the apparatus includes:

a reading module configured to read the exchange status table, wherein the exchange status table is an M row and an N column state table; wherein the row order of the first to M rows of the exchange state table indicates a different first level switching device The data of the first to the Nth columns of the exchange state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the congestion management device indicating the connection with the first-stage switching device. Link information, where M and N are positive integers;

a first selection module, configured to select all L first-level switching devices belonging to the rack J according to the exchange state table, and acquire K links corresponding to the L first-level switching devices; , L and J are positive integers less than M, and the K is a positive integer less than the product of L and N;

The first selection module is further configured to select L links from the K links;

a first accumulating module, configured to: after the first selection module selects L links from the K links, accumulating bandwidths of the L links successively, and accumulating results of two adjacent connections The first bandwidth accumulated value and the second bandwidth accumulated value are respectively included, wherein the first bandwidth accumulated value is a bandwidth accumulated value of the a links, and the second bandwidth accumulated value is a bandwidth accumulated value of the b links, The a link and the b link are all links in the L links, and the a and b are both positive integers less than or equal to L, and b=a+1;

The first shutdown module is configured to: when it is determined that the first bandwidth accumulation value is less than a preset bandwidth, and the second bandwidth accumulation value is greater than or equal to the preset bandwidth, shutting down the K links in the K links A link outside the link.

An embodiment of the present invention further provides a congestion management apparatus for a switching network, including: a memory and a processor connected to the memory; the memory stores computer executable instructions; and the processor is configured to execute the computer executable The instruction executes the congestion management method of the foregoing switching network.

The embodiment of the invention further provides a computer storage medium storing computer executable instructions, which can implement a congestion management method of a switched network after the computer executable instructions are executed.

The congestion management method and apparatus for the switching network and the computer storage medium provided by the embodiment of the present invention read the exchange state table, and the exchange state table is an M row and an N column state table; wherein the exchange state table is 1 to M The row order of the rows indicates the number of the different first-level switching devices, and the data of the first to Nth columns of the switching state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the indication Link information in the congestion management device connected to the first-stage switching device, wherein the M and N are positive integers; and all L first-level switching devices belonging to the rack J are selected according to the exchange state table, And acquiring K links corresponding to the L first-level switching devices; wherein, L and J are positive integers smaller than M, and the K is a positive integer smaller than a product of L and N; The L links are selected in the K links, and the bandwidths of the L links are successively accumulated, and the two adjacent accumulated results are respectively used as the first bandwidth accumulated value and the second bandwidth accumulated value; The first bandwidth accumulation value is a bandwidth accumulation value of a link, The second bandwidth accumulation value is a bandwidth accumulation value of the b links, and the a link and the b links are links in the L links, and the a and b are both less than or a positive integer equal to L, and b=a+1; determining that the first bandwidth accumulated value is less than a preset bandwidth, and the second bandwidth accumulated value is greater than or equal to the preset bandwidth, turning off the K chain A link other than the a link in the road. It can be seen that the embodiment of the present invention can evenly distribute the closed links on different first-level switching units, thereby solving the imbalance of the first-level switching unit links that are closed when the second-level switching units are asymmetric. The problem of congestion of the first-level switching unit ensures the traffic level of the entire network and improves the performance of the system.

DRAWINGS

1 is a schematic structural diagram of a three-stage CLOS switching network;

2 is a schematic structural diagram of a three-stage CLOS asymmetric switching network;

3 is a schematic flowchart of implementing a congestion management method for a switching network according to Embodiment 1 of the present invention;

4 is a schematic structural diagram of another three-level CLOS asymmetric switching network according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a congestion management apparatus according to Embodiment 1 of the present invention.

detailed description

The study found that asymmetric switching networks may cause congestion and packet loss. For example, when the asymmetry between the second-stage switching device and the third-level switching device, that is, the input end and the output bandwidth of the second-stage switching device do not match, the transmission link of the first-stage switching device is turned off to ensure The input and output bandwidths of the second-stage switching device are the same, thereby solving the problem of network congestion and packet loss caused by the consistent input and output bandwidth of the second-stage switching device. However, at this time, the second-level switching device of the tertiary network is asymmetric. The source bandwidth is the number of the chassis number. The number of links in the source chassis that are equivalent to the bandwidth loss of the destination switching access is closed. The link is likely to belong to the same first-level switching device; in the general networking structure, the second-level switching device is fully reachable, and it is impossible for all the links between the first-level switching device and the second-level switching device to be closed. The first-stage switching device after the link is closed is still reachable to the destination switching access. Therefore, the traffic of the source switching access is still equally distributed to each first-level switching device, and if a certain first level The plurality of transmission links of the switching device are turned off, and congestion will occur on the first-stage switching device. In the asymmetric network shown in FIG. 2, the third-stage switching device SU2# is asymmetric, and when the asymmetric processing is performed, the link corresponding to the central chassis 66 is closed, which may cause the second-level switching device SU3#. Or the asymmetry of SU4#. Here, taking the second-stage switching device SU3# as an example, the asymmetric result of the second-level switching device is processed, and the link between the service chassis 1 and the service chassis 2 is closed. When the link of the service chassis 1 is closed, the link number configured on the chassis table where the second-level switching device SU3# is located is continuously turned off. When different products are applied, the link distribution in the chassis table is If the first-stage switching devices are not evenly distributed, the corresponding first-level switching devices are different. In most scenarios, when the link of the service chassis 1 is closed, the links of the first-level switching devices SU1# and SU2# are not evenly distributed. In the extreme case, only the link of the first-stage switching device SU1# is turned off. When multiple links of the first-stage switching device SU1# are closed and not completely closed, the original switching access devices SAU1# and SAU2# of the service box 1 cannot perceive the first level connected thereto. If the links of the switching devices SU1# and SU2# change, they will still send the traffic evenly to the first-stage switching devices SU1# and SU2#, and the first-stage switching device SU1# will have more output links. When the strip is closed, the asymmetry of the first-stage switching device SU1# is caused, which causes the entire switching network to be congested or even lost. Therefore, in the embodiment of the present invention, a method for achieving a uniform link after the first-stage switching unit is closed is provided when the second-stage switching unit is asymmetric, so as to reduce network congestion and packet loss caused by the unbalanced link. The technical solutions in the embodiments of the present invention will be clearly and completely described in the following with reference to the accompanying drawings.

Embodiment 1

FIG. 3 is a schematic flowchart of a method for implementing congestion management in a switching network according to Embodiment 1 of the present invention. As shown in FIG. 3, the congestion management method in this embodiment includes the following steps:

Step 101: Read the exchange status table.

Here, Table 1 is an exchange state table, which may also be referred to as an extended rack table, wherein the exchange state table is an M row, an N column state table, and the switching state table is 1 to M rows. The row order indicates the number of the different first-level switching devices, and the data of the first to N-th columns of the exchange state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the representation and the The link information in the congestion management device connected to the first-level switching device, wherein M and N are positive integers.

As shown in Figure 4, the schematic diagram of a three-stage CLOS asymmetric switching network includes four service chassis, namely, service chassis 1 to service chassis 4, and two central chassis, namely, the central chassis 65 and the central chassis. 66, and each service chassis has 4 source switching access devices 401 and 4 first level switching devices 402, or 4 destination switching access devices 405 and 4 third level switching devices 404, and each There are four second-level switching devices 403 on the center chassis. In the embodiment of the present invention, the chassis may also be referred to as a chassis. In general, a switching device includes a plurality of switching units. For convenience of drawing and description, in FIG. 4, one switching unit can be regarded as one switching device, and one source switching access unit can be regarded as a source switching access device and A destination switching access unit can be considered as a destination switching access device. It is worth noting that, in order to distinguish, the switching unit connected to the source switching access unit on the service chassis is called the first-level switching unit, and the switching unit on the central chassis is called the second-level switching unit. The switching unit to which the destination switching access unit is connected is called a third-level switching unit. Correspondingly, the switching device that configures the first-level switching unit is the first-level switching device, and other devices are similar, and details are not described herein again.

Here, the asymmetric switching network refers to a situation in which bandwidths are inconsistent between two levels adjacent to the switching network, that is, the bandwidth of all the nodes in any one level is inconsistent with the other node.

Here, the congestion management device is a second-level switching device placed on the central chassis. In the embodiment of the present invention, a second-level switching device is present, which refers to the congestion management device.

Serial number	1	2	3	4	5	6	7	8	9	10	11	12	...	144	145
1	0	1	0	0	0	0	0	0	0	0	0	0	...	0	0
2	0	0	1	0	1	0	0	0	0	0	0	0	...	0	0
3	0	0	0	0	0	0	1	0	1	1	0	0	...	0	0
4	0	0	0	0	0	0	0	0	0	0	1	0	...	0	1

5	0	0	0	0	0	0	0	1	0	0	0	0	...	0	0
6	0	0	0	0	0	1	0	0	0	0	0	0	...	0	0
7	1	0	0	0	0	0	0	0	0	0	0	1	...	0	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
144	0	0	0	0	0	0	0	0	0	0	0	1	...	0	1

Table 1

Optionally, the method further includes: the Nth column data of the exchange state table is used to identify the rack identification information of the first level switching device, as shown in Table 1. For example, first, the row numbers 1 to 144 in the exchange state table represent the first-stage switching device numbers numbered 1 to 144; secondly, the data identifiers in the 145th column of the exchange state table are placed in the first-stage switching device. The rack identification information, for example, the data in column 145 shows the number "1", indicating that the first-stage switching device corresponding to the next row of the row of the number "1" is the first-stage switching device in the other rack. , the data in the 1st to 4th rows of the 145th column is {0, 0, 0, 1}, and the data in the 5th to 7th rows of the 145th column is {0, 0, 1}, indicating The first-stage switching devices corresponding to the 1st to 4th rows are placed in the same rack, and the first-stage switching devices corresponding to the 5th to 7th rows are in another rack; finally, the first to the first The data in column 144 represents link information in the congestion management device connected to the first-stage switching device, 0 represents no link connection in the first-level switching device and the congestion management device, and 1 represents the first The level switching device is connected to the congestion management device by a link. For example, the data in the first row and the first column is 0, indicating that the number is 1. Level switching means and congestion management means has link connections, the data row 1, column 2 is 1, a number of the first stage switching means and congestion management apparatus 1 is connected with a link. It is worth noting that the first row in Table 1 is the row with row number 1, and the first column is the column with column number 1.

In summary, the exchange status table includes: a number of the first-level switching device, rack identification information of the first-stage switching device, and link information in the congestion management device connected to the first-stage switching device, so that When the first problem of the second-level switching device is solved by the link of the switching-level switching device, the link balancing of one or more first-level switching devices can be closed according to the information in the switching state table. Therefore, the problem of frequent network congestion and packet loss caused by closed link imbalance is eliminated.

Optionally, before the reading the exchange state table, the method further includes: detecting whether a bandwidth of an input link in the congestion management device is greater than a bandwidth of an output link; and determining a bandwidth of the input link When the bandwidth of the output link is greater than, the exchange state table is read.

Optionally, in each service machine frame, two links are respectively connected between the switching access unit and the switching unit, and the switching unit on the service chassis and the switching unit on the central chassis are also respectively There are two links connected. At this time, there is a link disconnected between the switching unit 2# on the service chassis 3 and the destination switching access unit 1#. As shown in FIG. 4, the dotted arrow line indicates the link. If the disconnection occurs, then the switching network has a third-level switching unit and the destination switching access unit is asymmetric. On the service box 3, the input bandwidth of the third-pole switching unit 2# is larger than the output bandwidth, which may be generated here. Local congestion, resulting in system packet loss, the entire network traffic decreased.

When the third-level switching unit-destination switching access unit is asymmetric, the processing manner is to close the output link of the second-level switching unit (the central chassis 66), and after the output link of the second-level is closed, The second-stage switching unit has a problem that the input bandwidth is larger than the output bandwidth, which also causes congestion of the second-stage switching unit, and needs to be a non-pairing process of the second-stage switching device. At this time, the switching state is read. table.

Optionally, after the reading the exchange status table, the method further includes: performing a Leading Zero Detection (LZD) and rearranging the read data for the link information, and generating Link data set.

Step 102: Select all L first-level switching devices belonging to the rack J according to the exchange state table, and acquire K links corresponding to the L first-level switching devices.

Here, the L and J are positive integers smaller than M, and the K is a positive integer smaller than the product of L and N.

Optionally, after the selecting, according to the exchange state table, all the L first-level switching devices that belong to the rack J, the method further includes: obtaining, from the link data set, the belonging to the L Data of the first level switching device;

Merging the data attributed to the L first-stage switching devices;

And the acquiring the K links corresponding to the L first-level switching devices includes: acquiring K links corresponding to the L first-level switching devices according to the merged result.

Here, the data attributed to the L first-level switching devices refers to: L first numbers belonging to the same rack or chassis in the exchange state table because the row number indicates the first-level switching device The data corresponding to the level switching device, that is, the L row and N column data in the exchange state table, the row number of the L row indicates the number of the L first level switching devices.

Optionally, all L first-level switching devices belonging to the rack J are selected according to the exchange state table, and the first-level switching devices belonging to the rack J are corresponding to the exchange state table (ie, L rows and N columns). The information of the exchange state table is merged in rows, and K links corresponding to the L first-level switching devices are obtained, and information about the link is obtained. For example, as shown in Table 1, the information in the first row to the fourth row is merged to obtain rack-link information (rack_link), and rack_link={(z4, d11), (z3, d10) , (z3, d9), (z3, d7), (z2, d5), (z2, d3), (z1, d2)}, where (z4, d11) represents congestion management connected to the first-stage switching device 4. Link 11 in the device. After the leading zero detection, the number of links corresponding to the first-stage switching device LZDcnt is combined into {1, 3, 2, 1}, that is, one link corresponding to the first-stage switching device 4, and the first level There are three links corresponding to the switching device 3, two links corresponding to the first-stage switching device 2, and one link corresponding to the first-stage switching device 1. Here, z denotes a row of the exchange state table, and d denotes a column of the exchange state table.

Step 103: Select L links from the K links.

For example, for a link with 7 links, rack_link={(z4, d11), (z3, d10), (z3, d9), (z3, d7), (z2, d5), (z2, D3), (z1, d2)}, when selecting a link, select a link in the link corresponding to each first-level switching device, and obtain rack_link (z1, d2), rack_link (z2, d3), rack_link ( Z3, d7) and rack_link (z4, d11). In general, when a link is selected, a link with a smaller link bandwidth can be selected.

Step 104: Accumulate the bandwidths of the L links one by one, and use the two adjacent accumulated results as the first bandwidth accumulated value and the second bandwidth accumulated value, respectively.

Here, the first bandwidth accumulation value is a bandwidth accumulation value of a link, and the second bandwidth accumulation value is a bandwidth accumulation value of the b links, where the a link and the b link are both For the links in the L links, the a and b are both positive integers less than or equal to L, and b=a+1. For example, in the four links rack_link(z1,d2), rack_link(z2,d3), rack_link(z3,d7), and rack_link(z4,d11), two of the links rack_link(z1,d2) and rack_link are used. The bandwidths of (z2, d3) are added, and the first bandwidth accumulation result is L1, and the bandwidths of the three links rack_link (z1, d2), rack_link (z2, d3), and rack_link (z3, d7) are added. The second bandwidth accumulation result is obtained as L2.

Step 105: Determine the first bandwidth accumulated value, the second bandwidth accumulated value, and the size of the preset bandwidth.

Here, the method for determining the preset bandwidth is: polling all link states connected to other switching units by a switching unit, and accumulating bandwidths of all links, when all valid link polling ends, The accumulated result is the output bandwidth, and the output bandwidth is taken as the preset bandwidth value. It should be noted that the preset bandwidth is a congestion management device or other bandwidth test device can be determined by the above method, but is not limited to the above method.

Step 106: Close the links of the K links except the a links.

Optionally, when it is determined that the first bandwidth accumulated value is less than a preset bandwidth, and the second bandwidth accumulated value is greater than or equal to the preset bandwidth, turning off the a link in the K links A link other than that. For example, if the preset bandwidth is w, the bandwidth accumulation value L1 of the link rack_link (z1, d2) and rack_link (z2, d3) is less than w, rack_link (z1, d2), rack_link (z2, d3), and The bandwidth accumulation value L2 of rack_link(z3, d7) is greater than w. At this time, 7 links rack_link={(z4, d11), (z3, d10), (z3, d9), (z3, d7), (z2) , d5), (z2, d3), (z1, d2)}, keep the link rack_link(z1, d2) and rack_link(z2, d3), link link_link(z3, d7), rack_link(z3, d9) ), rack_link (z3, d10) and rack_link (z4, d11) are closed, thereby ensuring that the input bandwidth of the second set switching device is consistent with the output bandwidth, thereby avoiding network congestion.

Step 107: Determine whether b is equal to L.

When the first bandwidth accumulated value and the second bandwidth accumulated value are both smaller than the preset bandwidth, it is determined whether the bandwidth of the L links is accumulated, that is, whether b is equal to L, and when b is equal to L, the L links are The bandwidth has been accumulated; when b is less than L, it indicates that the L links have not been fully accumulated. At this time, let a=a+1, b=b+1, and then return to step 104, that is, the The bandwidths of the L links are successively accumulated, and the two adjacent accumulated results are respectively used as the first bandwidth accumulated value and the second bandwidth accumulated value, as shown in FIG. 3 .

Step 108: Select K-L links other than the L links from the K links.

Optionally, when it is determined that b is equal to L, a K-L link other than the L links is selected from the K links. For example, 7 links rack_link={(z4,d11), (z3,d10), (z3,d9), (z3,d7), (z2,d5),(z2,d3),(z1, In d2)}, four links of rack_link (z1, d2), rack_link (z2, d3), rack_link (z3, d7) and rack_link (z4, d11) have been selected, and the bandwidth of the four links is accumulated. Both are smaller than the preset bandwidth. Therefore, the four links are reserved, and the remaining three links are reacquired, that is, the link rack_link (z4, d11), rack_link (z3, d10), and rack_link (z3, d9). ).

Step 109: Determine whether K-L is greater than L.

Here, for each selection, generally one link is selected in the link corresponding to each first-level switching device, that is, L switching devices generally select L links. When K-L is greater than L, the process proceeds to step 110; when K-L is less than L, then the K-L links are all selected.

Step 110: Let K=K-L, and perform the next operation of selecting L links from the K links.

Optionally, when it is determined that KL is greater than L, the KL is taken as a new K value, and then returns to step 103, that is, the bandwidth of the L links is successively accumulated, and the two adjacent ones are adjacent. The accumulated result is used as the first bandwidth accumulated value and the second bandwidth accumulated value respectively, and the first bandwidth accumulated value and the second bandwidth accumulated value are respectively the bandwidth accumulated values of the a and b links, respectively.

Step 111: Accumulate the bandwidths of the K-L links one by one, and use the two adjacent accumulated results as the third bandwidth accumulated value and the fourth bandwidth accumulated value, respectively.

Optionally, when it is determined that K-L is less than or equal to L, the bandwidth of the K-L link is successively accumulated, and the two adjacent accumulated results are respectively used as the third bandwidth accumulated value and the fourth bandwidth accumulated value.

Here, the third bandwidth accumulation value is a bandwidth accumulation value of the c links, and the fourth bandwidth accumulation value is a bandwidth accumulation value of the d links, where the c links and the d links are both A link in a KL link, the c and d being positive integers less than or equal to KL, and d=c+1. For example, in the remaining three links rack_link (z4, d11), rack_link (z3, d10), and rack_link (z3, d9), the bandwidths of the three links are sequentially accumulated to obtain a link rack_link. The third bandwidth accumulated value L3 of (z4, d11) and rack_link (z3, d10), and the fourth bandwidth accumulated value L4 of the link rack_link (z4, d11), rack_link (z3, d10), and rack_link (z3, d9) .

Step 112: Determine whether the third bandwidth accumulated value is smaller than the preset bandwidth, and whether the fourth bandwidth accumulated value is greater than or equal to the preset bandwidth.

Step 113: Close links other than the c links except the K-L link.

Optionally, when it is determined that the third bandwidth accumulated value is smaller than the preset bandwidth, and the fourth bandwidth accumulated value is greater than or equal to the preset bandwidth, shutting down the KL link except the c A link outside the link. For example, if the preset bandwidth is w, if the third bandwidth accumulated value L3 of the link rack_link (z4, d11) and rack_link (z3, d10) is less than w, and the link rack_link (z4, d11), The fourth bandwidth accumulated value L4 of rack_link(z3, d10) and rack_link(z3, d9) is greater than w. At this time, rack_link(z4, d11), rack_link(z3, d10), and rack_link(z3, d9) are three links. The rack_link (z3, d9) is closed.

Step 114: Determine if d is equal to K-L.

When the third bandwidth accumulated value and the fourth bandwidth accumulated value are both smaller than the preset bandwidth, it is necessary to determine whether the bandwidth of the KL link is accumulated, that is, whether d is equal to L, and when d is equal to KL, the KL link is The bandwidth has been accumulated; when d is less than L, it indicates that the KL link has not been completely accumulated. At this time, let K=KL, and then need to return to step 111, that is, the bandwidth of the KL link is successively performed. Performing the accumulation, the two adjacent accumulated results are respectively used as the third bandwidth accumulated value and the fourth bandwidth accumulated value, and the first bandwidth accumulated value and the second bandwidth accumulated value are the bandwidth accumulated values of the a and b links, respectively. Step, the value of a here, as shown in Figure 3.

Step 115: Let J=J+1, and then perform the next operation of selecting all L first-level switching devices belonging to the rack J according to the exchange state table.

Determining that the d is equal to the KL, and the fourth bandwidth accumulated value is less than the preset bandwidth, adding J to the calculated value and taking the calculated sum value as a new J value, and then performing the next according to the The exchange status table selects the operations of all L first-level switching devices belonging to the rack J.

In order to implement the foregoing method, the first embodiment of the present invention further provides a congestion management apparatus. As shown in FIG. 5, the congestion management apparatus includes: a reading module 501, a first selection module 502, a first accumulation module 503, and a a shutdown module 504; wherein

The reading module 501 is configured to read the exchange status table, where the exchange status table is an M row and an N column state table; wherein the row order of the first to M rows of the exchange state table indicates different first level switching devices The data of the first to the Nth columns of the exchange state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the congestion management device indicating the connection with the first-stage switching device. Link information, where M and N are positive integers;

The first selection module 502 is configured to select all L first-level switching devices belonging to the rack J according to the exchange state table, and acquire K links corresponding to the L first-level switching devices; Wherein L and J are positive integers smaller than M, and the K is a positive integer smaller than a product of L and N;

The first selection module 502 is further configured to select L links from the K links;

The first accumulation module 503 is configured to, after the first selection module 502 selects L links from the K links, accumulate the bandwidths of the L links one by one, and add the two adjacent accumulation results. The first bandwidth accumulated value and the second bandwidth accumulated value are respectively included, wherein the first bandwidth accumulated value is a bandwidth accumulated value of the a links, and the second bandwidth accumulated value is a bandwidth accumulated value of the b links, The a link and the b link are all links in the L links, and the a and b are both positive integers less than or equal to L, and b=a+1;

The first closing module 504 is configured to: when it is determined that the first bandwidth accumulated value is less than a preset bandwidth, and the second bandwidth accumulated value is greater than or equal to the preset bandwidth, shutting down the K links except the a A link outside the link.

Optionally, the device further includes: a second selecting module 505, configured to: when the b is equal to the L, and the second bandwidth accumulated value is less than the preset bandwidth, from the K chain Selecting a KL link other than the L links in the path;

The determining module 506 is configured to determine whether the K-L is greater than the L;

The second accumulating module 507 is configured to accumulate the bandwidth of the KL link one by one when the determining module 506 determines that the KL is less than or equal to the L, and respectively use the two adjacent accumulated results as the third a bandwidth accumulated value and a fourth bandwidth accumulated value; wherein the third bandwidth accumulated value is a bandwidth accumulated value of the c links, and the fourth bandwidth accumulated value is a bandwidth accumulated value of the d links, the c pieces The link and the d link are links in the KL link, and the c and d are positive integers less than or equal to KL, and d=c+1;

The second closing module 508 is configured to: when it is determined that the third bandwidth accumulated value is less than the preset bandwidth, and the fourth bandwidth accumulated value is greater than or equal to the preset bandwidth, shutting down the KL link a link other than the c links;

The first processing module 509 is configured to: when the d is equal to the KL, and the fourth bandwidth accumulated value is less than the preset bandwidth, add one to the J, and use the calculated sum value as a new one. J value, and the next operation of selecting all L first-level switching devices belonging to the rack J according to the exchange state table;

The second processing module 510 is configured to: when the determining module determines that the KL is greater than the L, use the KL as a new K value, and perform the next time selecting L links from the K links. Operation.

Optionally, the Nth column data of the exchange state table is used to identify the rack identification information of the first level switching device.

Optionally, the device further includes: a detecting module 511, configured to detect whether a bandwidth of the input link in the congestion management device is greater than a bandwidth of the output link;

The reading module 501 is further configured to read the exchange state table when determining that the bandwidth of the input link is greater than the bandwidth of the output link.

Optionally, the device further includes: a third processing module 512, configured to perform pre-zero detection on the read data for the link information after the reading module reads the exchange state table LZD, and rearrange, to generate a link data set.

Optionally, the device further includes: a third selecting module 513, configured to: after the first selecting module 502 selects all the L first-level switching devices belonging to the rack J according to the switching state table, Obtaining data belonging to the L first-level switching devices in the link data set;

The merging module 514 is configured to merge the data attributed to the L first-level switching devices;

The first selection module 502 is specifically configured to acquire K links corresponding to the L first-level switching devices according to the merged result.

The congestion management device composed of the above structure can perform the following method steps:

(1) The reading module 501 reads the exchange status table.

Here, Table 1 is an exchange state table, which may also be referred to as an extended rack table, wherein the exchange state table is an M row, an N column state table, and the switching state table is 1 to M rows. The row order indicates the number of the different first-level switching devices, and the data of the first to N-th columns of the exchange state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the representation and the The link information in the congestion management device connected to the first-level switching device, wherein M and N are positive integers.

As shown in Figure 4, the schematic diagram of a three-stage CLOS asymmetric switching network includes four service chassis, namely, service chassis 1 to service chassis 4, and two central chassis, namely, the central chassis 65 and the central chassis. 66, and each service chassis has 4 source switching access devices 401 and 4 first level switching devices 402, or 4 destination switching access devices 405 and 4 third level switching devices 404, and each There are four second-level switching devices 403 on the center chassis. In the embodiment of the present invention, the chassis may also be referred to as a chassis. In general, a switching device includes a plurality of switching units. For convenience of drawing and description, in FIG. 4, one switching unit can be regarded as one switching device, and one source switching access unit can be regarded as a source switching access device and A destination switching access unit can be considered as a destination switching access device. It is worth noting that, in order to distinguish, the switching unit connected to the source switching access unit on the service chassis is called the first-level switching unit, and the switching unit on the central chassis is called the second-level switching unit. The switching unit to which the destination switching access unit is connected is called a third-level switching unit. Correspondingly, the switching device that configures the first-level switching unit is the first-level switching device, and other devices are similar, and details are not described herein again.

Here, the asymmetric switching network refers to a situation in which bandwidths are inconsistent between two levels adjacent to the switching network, that is, the bandwidth of all the nodes in any one level is inconsistent with the other node.

Here, the congestion management device is a second-level switching device placed on the central chassis. In the embodiment of the present invention, a second-level switching device is present, which refers to the congestion management device.

Serial number	1	2	3	4	5	6	7	8	9	10	11	12	...	144	145
1	0	1	0	0	0	0	0	0	0	0	0	0	...	0	0
2	0	0	1	0	1	0	0	0	0	0	0	0	...	0	0
3	0	0	0	0	0	0	1	0	1	1	0	0	...	0	0

4	0	0	0	0	0	0	0	0	0	0	1	0	...	0	1
5	0	0	0	0	0	0	0	1	0	0	0	0	...	0	0
6	0	0	0	0	0	1	0	0	0	0	0	0	...	0	0
7	1	0	0	0	0	0	0	0	0	0	0	1	...	0	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
144	0	0	0	0	0	0	0	0	0	0	0	1	...	0	1

Table 1

Optionally, the method further includes: the Nth column data of the exchange state table is used to identify the rack identification information of the first level switching device, as shown in Table 1. For example, first, the row numbers 1 to 144 in the exchange state table represent the first-stage switching device numbers numbered 1 to 144; secondly, the data identifiers in the 145th column of the exchange state table are placed in the first-stage switching device. The rack identification information, for example, the data in column 145 shows the number "1", indicating that the first-stage switching device corresponding to the next row of the row of the number "1" is the first-stage switching device in the other rack. , the data in the 1st to 4th rows of the 145th column is {0, 0, 0, 1}, and the data in the 5th to 7th rows of the 145th column is {0, 0, 1}, indicating The first-stage switching devices corresponding to the 1st to 4th rows are placed in the same rack, and the first-stage switching devices corresponding to the 5th to 7th rows are in another rack; finally, the first to the first The data in column 144 represents link information in the congestion management device connected to the first-stage switching device, 0 represents no link connection in the first-level switching device and the congestion management device, and 1 represents the first The level switching device is connected to the congestion management device by a link. For example, the data in the first row and the first column is 0, indicating that the number is 1. Level switching means and congestion management means has link connections, the data row 1, column 2 is 1, a number of the first stage switching means and congestion management apparatus 1 is connected with a link. It is worth noting that the first row in Table 1 is the row with row number 1, and the first column is the column with column number 1.

Optionally, before the reading module 501 reads the exchange state table, the method further includes: detecting, by the detecting module 511, whether a bandwidth of the input link in the congestion management device is greater than a bandwidth of the output link; When the bandwidth of the input link is greater than the bandwidth of the output link, the reading module 501 reads the exchange status table.

Optionally, in each service machine frame, two links are respectively connected between the switching access unit and the switching unit, and the switching unit on the service chassis and the switching unit on the central chassis are also respectively There are two links connected. At this time, there is a link disconnected between the switching unit 2# on the service chassis 3 and the destination switching access unit 1#. As shown in FIG. 4, the dotted arrow line indicates the link. Disconnected, then the third-level switching unit-destination switching access unit is asymmetric in the switching network. On the service box 3, the input bandwidth of the third-pole switching unit 2# is larger than the output bandwidth, which may be generated here. Local congestion, resulting in system packet loss, the entire network traffic decreased.

When the third-level switching unit-destination switching access unit is asymmetric, the processing manner is to close the output link of the second-level switching unit (the central chassis 66), and after the output link of the second-level is closed, The second-stage switching unit has a problem that the input bandwidth is larger than the output bandwidth, which also causes congestion of the second-level switching unit, and needs to be processed by the second-stage switching device. At this time, the reading module 501 reads The exchange status table.

Optionally, after the reading module 501 reads the exchange status table, the method further includes: the third processing module 512 performs LZD and rearranges the read data for the link information to generate a link. data set.

(2) The first selection module 502 selects all the L first-level switching devices belonging to the rack J according to the exchange state table, and acquires K links corresponding to the L first-level switching devices.

Here, the L and J are positive integers smaller than M, and the K is a positive integer smaller than the product of L and N.

Optionally, after the first selection module 502 selects all the L first-level switching devices that belong to the chassis J according to the exchange state table, the method further includes: the third selection module 513 from the link data. Collecting data belonging to the L first-level switching devices in a centralized manner;

The merging module 514 merges the data attributed to the L first-level switching devices;

The first selection module 502 acquires the K links corresponding to the L first-level switching devices, and the first selection module 502 acquires the K-chains corresponding to the L first-level switching devices according to the merged result. road.

Here, the data attributed to the L first-level switching devices refers to: L first numbers belonging to the same rack or chassis in the exchange state table because the row number indicates the first-level switching device The data corresponding to the level switching device, that is, the L row and N column data in the exchange state table, the row number of the L row indicates the number of the L first level switching devices.

Optionally, the third selecting module 513 selects all L first-level switching devices belonging to the rack J according to the switching state table, and the combining module 514 exchanges the first-level switching devices belonging to the rack J. The information of the state table (ie, the L row and the N column exchange state table) is merged by the row, and the first selection module 502 acquires K links corresponding to the L first-level switching devices to obtain information about the link. For example, as shown in Table 1, the information in the first row to the fourth row is summed to obtain rack_link={(z4, d11), (z3, d10), (z3, d9), (z3, d7). (z2, d5), (z2, d3), (z1, d2)}, where (z4, d11) represents the 11th link in the congestion management apparatus connected to the first-stage switching device 4. After performing the leading zero detection, the number of links corresponding to the first-stage switching device LZDcnt is combined into {1, 3, 2, 1}, that is, one link corresponding to the first-stage switching device 4, and the first The level switching device 3 has three links, two links corresponding to the first-stage switching device 2, and one link corresponding to the first-stage switching device 1. Here, z denotes a row of the exchange state table, and d denotes a column of the exchange state table.

(3) The first selection module 502 selects L links from the K links.

For example, for a link with 7 links, rack_link={(z4, d11), (z3, d10), (z3, d9), (z3, d7), (z2, d5), (z2, D3), (z1, d2)}, when the first selection module 502 selects a link, selects one link in the link corresponding to each first-level switching device, and obtains rack_link (z1, d2) and rack_link (z2, D3), rack_link (z3, d7) and rack_link (z4, d11). In general, when the first selection module 502 selects a link, a link with a smaller link bandwidth is preferred.

(4) The first accumulation module 503 accumulates the bandwidths of the L links one by one, and uses the two adjacent accumulation results as the first bandwidth accumulated value and the second bandwidth accumulated value, respectively.

Here, the first bandwidth accumulation value is a bandwidth accumulation value of a link, and the second bandwidth accumulation value is a bandwidth accumulation value of the b links, where the a link and the b link are both For the links in the L links, the a and b are both positive integers less than or equal to L, and b=a+1. For example, in the four links rack_link (z1, d2), rack_link (z2, d3), rack_link (z3, d7), and rack_link (z4, d11), the first accumulation module 503 will have two links rack_link (z1). , d2) and the link rack_link (z2, d3) bandwidth are added, the first bandwidth accumulation result is L1, three links rack_link (z1, d2), link rack_link (z2, d3) and link rack_link The bandwidths of (z3, d7) are added, and the second bandwidth accumulation result is L2.

(5) The determining module 506 determines the first bandwidth accumulated value, the second bandwidth accumulated value, and the size of the preset bandwidth.

Here, the method for determining the preset bandwidth is: polling all link states connected to other switching units by a switching unit, and accumulating bandwidths of all links, when all valid link polling ends, The accumulated result is the output bandwidth, and the output bandwidth is taken as the preset bandwidth value. It should be noted that the preset bandwidth is determined by the above method by a congestion management device or other bandwidth test device.

(6) The first shutdown module 504 closes links other than the a links among the K links.

Optionally, when the determining module 506 determines that the first bandwidth accumulated value is less than a preset bandwidth, and the second bandwidth accumulated value is greater than or equal to the preset bandwidth, the first closing module 504 closes the K links. A link other than the a link. For example, if the preset bandwidth is w, the bandwidth accumulation value L1 of the link rack_link (z1, d2) and rack_link (z2, d3) is less than w, and the link rack_link (z1, d2), rack_link (z2, D3) and rack_link (z3, d7) bandwidth accumulation value L2 is greater than w, at this time, 7 links rack_link = {(z4, d11), (z3, d10), (z3, d9), (z3, d7) , (z2, d5), (z2, d3), (z1, d2)}, retain the link rack_link (z1, d2) and rack_link (z2, d3), the link rack_link (z3, d7), rack_link ( Z3, d9), rack_link (z3, d10) and rack_link (z4, d11) are closed, thereby ensuring that the input bandwidth of the second set switching device is consistent with the output bandwidth, thereby avoiding network congestion.

(7) The decision module 506 determines whether b is equal to L.

When the first bandwidth accumulated value and the second bandwidth accumulated value are both smaller than the preset bandwidth, it is determined whether the bandwidth of the L links is accumulated, that is, whether b is equal to L, and when b is equal to L, the L links are The bandwidth has been accumulated; when b is less than L, it indicates that the L links have not been fully accumulated. At this time, let a=a+1, b=b+1, and then return to the execution step (4), that is, the The bandwidths of the L links are successively accumulated, and the two adjacent accumulated results are respectively used as the first bandwidth accumulated value and the second bandwidth accumulated value, as shown in FIG. 3 .

(8) The second selection module 505 selects K-L links other than the L links from the K links.

Optionally, when the determining module 506 determines that b is equal to L, the second selecting module 505 selects K-L links other than the L links from the K links. For example, 7 links rack_link={(z4,d11), (z3,d10), (z3,d9), (z3,d7), (z2,d5),(z2,d3),(z1, In d2)}, four links of rack_link (z1, d2), rack_link (z2, d3), rack_link (z3, d7) and rack_link (z4, d11) have been selected, and the bandwidth of the four links is accumulated. Both are smaller than the preset bandwidth. Therefore, the four links are reserved, and the remaining three links are reacquired, that is, the link rack_link (z4, d11), rack_link (z3, d10), and rack_link (z3, d9). ).

(9) The judgment module 506 determines whether K-L is greater than L.

Here, for each selection, generally one link is selected in the link corresponding to each first-level switching device, that is, L switching devices generally select L links. When K-L is greater than L, the process proceeds to step (10); when K-L is less than L, then the K-L links are all selected.

(10) The second processing module 510 makes K = K-L and performs the next operation of selecting L links from the K links.

Optionally, when the determining module 506 determines that KL is greater than L, the second processing module 510 takes the KL as a new K value, and then returns to step (3), that is, the bandwidth of the L links. The accumulating is performed successively, and the two adjacent accumulated results are respectively used as the first bandwidth accumulated value and the second bandwidth accumulated value, and the first bandwidth accumulated value and the second bandwidth accumulated value are respectively the bandwidth accumulated values of the a and b links respectively. step

(11) The second accumulation module 507 accumulates the bandwidths of the K-L links one by one, and uses the two adjacent accumulation results as the third bandwidth accumulated value and the fourth bandwidth accumulated value, respectively.

Optionally, when the determining module 506 determines that the KL is less than or equal to L, the second accumulating module 507 sequentially accumulates the bandwidth of the KL link, and uses the two adjacent accumulated results as the third bandwidth accumulated value. And the fourth bandwidth accumulated value.

Here, the third bandwidth accumulation value is a bandwidth accumulation value of the c links, and the fourth bandwidth accumulation value is a bandwidth accumulation value of the d links, where the c links and the d links are both A link in a KL link, the c and d being positive integers less than or equal to KL, and d=c+1. For example, in the remaining three links rack_link (z4, d11), rack_link (z3, d10), and rack_link (z3, d9), the bandwidths of the three links are sequentially accumulated to obtain a link rack_link. The third bandwidth accumulated value L3 of (z4, d11) and rack_link (z3, d10), and the fourth bandwidth accumulated value L4 of the link rack_link (z4, d11), rack_link (z3, d10), and rack_link (z3, d9) .

(12) The determining module 506 determines whether the third bandwidth accumulated value is smaller than the preset bandwidth, and whether the fourth bandwidth accumulated value is greater than or equal to the preset bandwidth.

(13) The second shutdown module 508 turns off links other than the c links in the K-L link.

Optionally, when it is determined that the third bandwidth accumulated value is smaller than the preset bandwidth, and the fourth bandwidth accumulated value is greater than or equal to the preset bandwidth, the second closing module 508 turns off the KL link. A link other than the c links. For example, if the preset bandwidth is w, if the third bandwidth accumulated value L3 of the link rack_link (z4, d11) and rack_link (z3, d10) is less than w, and the link rack_link (z4, d11), The fourth bandwidth accumulated value L4 of rack_link(z3, d10) and rack_link(z3, d9) is greater than w. At this time, the second closing module 508 will rack_link (z4, d11), rack_link (z3, d10) and the three links. The rack_link(z3, d9) of rack_link(z3, d9) is closed.

(14) The decision module 506 determines if d is equal to K-L.

When the third bandwidth accumulated value and the fourth bandwidth accumulated value are both smaller than the preset bandwidth, it is necessary to determine whether the bandwidth of the KL link is accumulated, that is, whether d is equal to L, and when d is equal to KL, the KL link is The bandwidth has been accumulated; when d is less than L, it indicates that the KL link has not been fully accumulated. At this time, let K=KL, and then need to return to the execution step (11), that is, the KL link is The bandwidth is successively accumulated, and the two adjacent accumulated results are respectively used as the third bandwidth accumulated value and the fourth bandwidth accumulated value, and the first bandwidth accumulated value and the second bandwidth accumulated value are the bandwidth accumulated values of the a and b links, respectively. This step, the value of a here, is shown in Figure 3.

(15) The first processing module 509 makes J=J+1, and then performs the next operation of selecting all of the L first-stage switching devices belonging to the rack J according to the exchange state table.

When it is determined that the d is equal to the KL, and the fourth bandwidth accumulated value is less than the preset bandwidth, the first processing module 509 increments J, takes the calculated sum value as a new J value, and then performs The operation of all L first-stage switching devices belonging to the rack J is selected next according to the exchange state table.

With the solution of the embodiment of the present invention, when a certain second-level switching device is asymmetric, the link connection relationship between the second-level switching device and the first-level switching device can be determined by searching the switching state table, and then The bandwidth of the corresponding link is selected and accumulated. When the bandwidth accumulated value is just equal to or greater than the threshold bandwidth value, the link that has been accumulated currently is retained, and other links are closed. Therefore, it is ensured that the closed links are evenly distributed on different first-level switching devices, thereby solving the imbalance of the first-level switching device links that are closed when the second-level switching devices are asymmetric, resulting in the first-level switching. The congestion of the device ensures the traffic level of the entire network, and also ensures that no bandwidth is wasted and the performance of the system is improved.

In a practical application, the reading module 501, the first selecting module 502, the first accumulating module 503, the first closing module 504, the second selecting module 505, the judging module 506, and the second accumulating module 507 of the congestion management apparatus, The second shutdown module 508, the first processing module 509, the second processing module 510, the detection module 511, the third processing module 512, the third selection module 513, and the merging module 514 may each be a central processing unit (CPU) located in the congestion management device. ), microprocessor (MPU), digital signal processor (DSP), or field programmable gate array (FPGA) implementation.

The embodiment further provides an embodiment of the present invention, further, a congestion management apparatus for a switching network, including: a memory and a processor connected to the memory; the memory stores computer executable instructions; and the processor is configured to execute The computer executable instructions perform one or more of the aforementioned congestion management methods of the switched network, for example, the method illustrated in FIG. 3 may be performed.

The memory can be various types of memory, such as various memories such as random access memory, read only memory, flash memory, and the like. The processor may be connected to a memory through a bus such as a computer integrated bus, and the processor may be a CPU, an MPU, a DSP, an FPGA, or the like.

The embodiment of the invention further provides a computer storage medium storing computer executable instructions, which can implement a congestion management method of a switched network after the computer executable instructions are executed.

The storage medium may be any type of storage medium, optionally a non-transitory storage medium such as a flash memory or a read only memory, an optical disk or a USB flash drive, or the like.

The above is only the preferred embodiment of the present invention and is not intended to limit the scope of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and scope of the present invention are included in the scope of the present invention.

Industrial applicability

The technical solution provided by the embodiment of the present invention, when the non-symmetric problem of the non-switched network is solved by closing the link of the first-level switching device, according to the identifier of the first-level switching device in the exchange state table, The frame identification information and the link information of the connection are selected by the bandwidth calculation, so that the bandwidth accumulation value of the corresponding link can be selected according to the link connection relationship between the second-level switching device and the first-level switching device. When the value is equal to or greater than the threshold bandwidth, the link that has been accumulated is retained, and the other links are closed, thereby ensuring that the closed links are evenly distributed on different first-level switching devices, thereby solving the second When the level switching device is asymmetric, the link of the first-level switching device that is closed is unbalanced, causing congestion of the first-level switching device, ensuring the traffic level of the entire network, and ensuring that no bandwidth is wasted, and the system is improved. performance. Therefore, the technical solution provided by the embodiment of the invention has a positive industrial effect, and at the same time has the characteristics of being simple and convenient, so that it can be widely used in industry.

Claims (14)

1. A congestion management method for a switching network, the method comprising:

   Reading the exchange state table, the exchange state table is an M row, N column state table; wherein the row order of the first to M rows of the exchange state table indicates the number of the different first level switching device, the switching state The data of the first to Nth columns of the table includes at least rack identification information indicating that the first-stage switching device is placed, and link information indicating a congestion management device connected to the first-stage switching device, M and N are both positive integers;

   Selecting, according to the exchange state table, all L first-level switching devices belonging to the rack J, and acquiring K links corresponding to the L first-level switching devices; wherein, L and J are a positive integer less than M, the K being a positive integer less than the product of L and N;

   Selecting L links from the K links, and accumulating the bandwidths of the L links one by one, and using the two adjacent accumulated results as the first bandwidth accumulated value and the second bandwidth accumulated value respectively; The first bandwidth accumulation value is a bandwidth accumulation value of a link, and the second bandwidth accumulation value is a bandwidth accumulation value of the b links, where the a link and the b link are both a link in the L links, wherein a and b are both positive integers less than or equal to L, and b=a+1;

   When it is determined that the first bandwidth accumulated value is less than a preset bandwidth, and the second bandwidth accumulated value is greater than or equal to the preset bandwidth, closing a chain other than the a links in the K links road.
2. The method of claim 1 wherein the method further comprises:

   Determining, when the b is equal to the L, and the second bandwidth accumulated value is less than the preset bandwidth, selecting a K-L link other than the L links from the K links;

   Determining whether the K-L is greater than the L;

   If the KL is less than or equal to the L, the bandwidth of the KL link is successively accumulated, and the two adjacent accumulated results are respectively used as the third bandwidth accumulated value and the fourth bandwidth accumulated value; The third bandwidth accumulating value is a bandwidth accumulating value of the c links, and the fourth bandwidth accumulating value is a bandwidth accumulating value of the d links, and the c links and the d links are all the KL strips. a link in the link, the c and d being positive integers less than or equal to KL, and d=c+1;

   When it is determined that the third bandwidth accumulated value is smaller than the preset bandwidth, and the fourth bandwidth accumulated value is greater than or equal to the preset bandwidth, turning off the c link except the KL link Outer link

   Determining that the d is equal to the KL, and the fourth bandwidth accumulated value is less than the preset bandwidth, adding one to the J, using the calculated sum value as a new J value, and performing the next basis The exchange state table selects operations of all L first-level switching devices belonging to the rack J;

   If the K-L is greater than the L, the K-L is taken as a new K value, and the next operation of selecting L links from the K links is performed.
3. The method of claim 2, wherein the method further comprises: the Nth column of data of the exchange state table is used to identify rack identification information for placing the first level switching device.
4. The method according to any one of claims 1 to 3, wherein the method further comprises:

   Detecting whether a bandwidth of the input link in the congestion management device is greater than a bandwidth of the output link;

   The exchange state table is read when it is determined that the bandwidth of the input link is greater than the bandwidth of the output link.
5. The method of claim 4, wherein after the reading the exchange status table, the method further comprises:

   The read data for the link information is subjected to leading zero detection LZD and rearranged to generate a link data set.
6. The method of claim 5, wherein the method further comprises: after selecting all of the L first-level switching devices belonging to the rack J according to the exchange state table, the method further comprising:

   Obtaining data belonging to the L first-level switching devices from the link data set;

   Merging the data attributed to the L first-stage switching devices;

   Obtaining K links corresponding to the L first-level switching devices, including:

   Obtaining K links corresponding to the L first-level switching devices according to the merged result.
7. A congestion management device for a switching network, the device comprising:

   a reading module configured to read the exchange status table, wherein the exchange status table is an M row and an N column state table; wherein the row order of the first to M rows of the exchange state table indicates a different first level switching device The data of the first to the Nth columns of the exchange state table includes at least the rack identification information indicating the placement of the first-stage switching device, and the congestion management device indicating the connection with the first-stage switching device. Link information, where M and N are positive integers;

   a first selection module, configured to select all L first-level switching devices belonging to the rack J according to the exchange state table, and acquire K links corresponding to the L first-level switching devices; , L and J are positive integers less than M, and the K is a positive integer less than the product of L and N;

   The first selection module is further configured to select L links from the K links;

   a first accumulating module, configured to: after the first selection module selects L links from the K links, accumulating bandwidths of the L links successively, and accumulating results of two adjacent connections The first bandwidth accumulated value and the second bandwidth accumulated value are respectively included, wherein the first bandwidth accumulated value is a bandwidth accumulated value of the a links, and the second bandwidth accumulated value is a bandwidth accumulated value of the b links, The a link and the b link are all links in the L links, and the a and b are both positive integers less than or equal to L, and b=a+1;

   The first shutdown module is configured to: when it is determined that the first bandwidth accumulation value is less than a preset bandwidth, and the second bandwidth accumulation value is greater than or equal to the preset bandwidth, shutting down the K links in the K links A link outside the link.
8. The apparatus of claim 7 wherein said apparatus further comprises:

   a second selection module, configured to: after determining that the b is equal to the L, and the second bandwidth accumulated value is less than the preset bandwidth, select the L links from the K links External KL link;

   a determining module, configured to determine whether the K-L is greater than the L;

   The second accumulating module is configured to accumulate the bandwidth of the KL link one by one when the determining module determines that the KL is less than or equal to the L, and use the two adjacent accumulated results as the third a bandwidth accumulated value and a fourth bandwidth accumulated value; wherein the third bandwidth accumulated value is a bandwidth accumulated value of the c links, and the fourth bandwidth accumulated value is a bandwidth accumulated value of the d links, the c pieces The link and the d link are links in the KL link, and the c and d are positive integers less than or equal to KL, and d=c+1;

   a second shutdown module, configured to: when it is determined that the third bandwidth accumulation value is less than the preset bandwidth, and the fourth bandwidth accumulation value is greater than or equal to the preset bandwidth, shutting down the KL link a link other than the c links;

   a first processing module, configured to: when determining that d is equal to the KL, and the fourth bandwidth accumulated value is less than the preset bandwidth, add one to the J, and use the calculated sum value as a new J Value, and the next operation of selecting all L first-level switching devices belonging to the rack J according to the exchange state table;

   a second processing module, configured to: when the determining module determines that the KL is greater than the L, the KL is used as a new K value, and the next time the L links are selected from the K links operating.
9. The apparatus of claim 8, wherein the column N data of the exchange state table is used to identify rack identification information for placing the first level switching device.
10. The device according to any one of claims 7 to 9, wherein the device further comprises:

    a detecting module, configured to detect whether a bandwidth of the input link in the congestion management device is greater than a bandwidth of the output link;

    The reading module is further configured to read the exchange status table when determining that the bandwidth of the input link is greater than the bandwidth of the output link.
11. The device of claim 10, wherein the device further comprises:

    The third processing module is configured to: after the reading module reads the exchange state table, perform the leading zero detection LZD on the read data for the link information, and rearrange the data to generate a link data set.
12. The apparatus of claim 11 wherein said apparatus further comprises:

    a third selection module, configured to: after the first selection module selects all the L first-level switching devices belonging to the rack J according to the exchange state table, obtain the attribution from the link data set Data of L first-stage switching devices;

    a merging module configured to merge the data attributed to the L first-level switching devices;

    The first selection module is configured to acquire K links corresponding to the L first-level switching devices according to the merged result.
13. A congestion management apparatus for a switching network, comprising: a memory and a processor coupled to the memory; the memory storing computer executable instructions; the processor being configured to execute claim 1 by executing the computer executable instructions To any of the methods provided in 6.
14. A computer storage medium storing computer executable instructions capable of implementing the method of any one of claims 1 to 6 after the computer executable instructions are executed.

Images