WO2016086385A1

WO2016086385A1 - Congestion control method, device and system

Info

Publication number: WO2016086385A1
Application number: PCT/CN2014/093015
Authority: WO
Inventors: 肖俊俊; 梁剑耀; 文白林
Original assignee: 华为技术有限公司
Priority date: 2014-12-04
Filing date: 2014-12-04
Publication date: 2016-06-09
Also published as: CN106134138A; CN106134138B

Abstract

Disclosed in an embodiment of the present invention is a congestion control method, comprising: a first network node conducts a priority-based flow control (PFC) detection for an inlet port, and if a data flow occurring congestion within a set detection period is detected, a flow identification of the data flow is obtained; the first network node generates a PFC frame comprising the flow identification of the data flow, and transmits the PFC frame to a second network node, so as to instruct the second network node to conduct the congestion control for the data flow. The embodiment of the present invention enables the PFC mechanism to conduct the congestion detection for the specific data flow by improving the existing PFC mechanism, transmits the congestion level to an uplink node via the improved PFC frame, and flexibly processes the data flow according to a congestion situation, thus effectively handling with the problems of the change of network data traffic and congestion diffusion, and improving efficiency of the whole network system. Also disclosed in the embodiment of the present invention are a network node, system and computer readable medium.

Description

Method, device and system for congestion control

Technical field

The embodiments of the present invention relate to the field of networks, and in particular, to a method, device, and system for congestion control.

Background technique

Ethernet (Ethernet) is a computer LAN networking technology. The IEEE 802.3 standard developed by the IEEE gives the technical standard for Ethernet, which specifies the contents of the connection, electrical signals, and medium access layer protocols including the physical layer. Ethernet is the most common LAN technology in current applications.

To solve the packet loss caused by congestion in Ethernet, a number of technologies for securing lossless Ethernet have been developed, including Quantized Congestion Notification (QCN) protocol and priority-based flow. Priority-based Flow Control (PFC) mechanism.

FIG. 1 is a schematic diagram of the working principle of the QCN mechanism in the prior art. The QCN protocol uses quantitative congestion control technology and uses a relatively accurate backward congestion notification mechanism compared to other congestion control technologies. The devices in the network are also referred to as network nodes; the congestion detection points are set on the network nodes based on the outbound port queues, that is, the congestion detection is performed on the cache queues that send out data on the network nodes. If the network node detects that congestion occurs, a congestion notification message (CNM) is generated, which includes information indicating the identification information of the flow causing the congestion and the degree of congestion; the CNM generated by the network node The message is sent along the reverse direction of the streaming to the source terminal that causes congestion, and the source terminal reduces the data transmission rate of the corresponding stream according to the CNM message.

2 is a schematic diagram of the working principle of the PFC mechanism in the prior art. A node that supports the PFC mechanism divides the data stream on the inbound port queue into eight priority queues according to the priority field of the VLAN tag specified by the 802.1Q protocol. Once a priority queue is detected (such as the priority queue 3 in Figure 2) If the excessive traffic causes congestion, the receiving end pauses the data reception on the priority queue 3, and the transmitting end stops the data transmission on the priority queue 3.

In the actual network, the QCN protocol and the PFC mechanism are generally used together, as shown in FIG. Network node E supports QCN protocol and PFC mechanism; A and D, B and D, D and E, C and E, and F and E use PFC mechanism for flow control. The thresholds for triggering the PFC mechanism and triggering the QCN mechanism are set on each node.

For example, on node E, the outbound port queue buffer size is 500K bytes, and the ingress port queue buffer size is 100K bytes. On node E, the threshold for triggering the QCN mechanism (that is, the flow rate of the rate at which the outbound port sends data) can be set to 20%, that is, once the outbound port cache occupancy exceeds 20% (that is, exceeds 100 kbytes), node E The QCN mechanism is activated to control the rate at which the outgoing port sends data.

The PFC machine is configured to be used for the ingress port. The threshold for triggering the PFC mechanism on the node E (that is, the flow rate control of the rate at which the inbound port receives data) can be set to 80%, that is, once the inbound port cache occupancy rate exceeds 80% (that is, 80K bytes), node E will start the PFC mechanism to control the rate at which the inbound port receives data. If the node E detects congestion caused by too much traffic of a certain priority queue as in FIG. 2, the data reception on the priority queue is suspended.

When the threshold of the triggered PFC mechanism set on the node E is too low, the node E can easily trigger the PFC mechanism, thereby causing the node E to reduce the rate at which the inbound port receives data, that is, the nodes C, D, and F go to the port to the node. E The rate at which data is sent. In this case, the egress port of the node E may never be congested, and the flow control function of the QCN mechanism on it will not be started; thus, the link utilization between the nodes is too low.

If the node E is congested for a long time, so that the ingress port queue on the node E is temporarily used 100%, the node E requests the node D, F or C to reduce the rate of transmitting data. For example, the node E requests the node D to reduce the rate at which the node D sends data to the node E. Once the rate at which the node D sends data to the node E is lowered, but the rate at which the inbound port of the temporary node D receives data cannot be lowered, The congestion caused by the ingress of the node D can be regarded as the fact that the node E spreads the congestion of the ingress port to the node D; and may continue to spread to the upper level node such as the node A or the node B.

When the threshold of the triggered PFC mechanism set on the node E is too high, the QCN mechanism on the node is triggered first, so that the rate at which the source terminal of the corresponding data stream transmits data decreases; and at this time, the PFC is triggered on the node E. If the threshold of the mechanism has not been reached, node E does not start the flow control function of the PFC mechanism. After the node E starts the QCN mechanism, if congestion is detected, a CNM message is generated and sent to the source terminal that causes congestion. However, the process needs to pass through multiple nodes in the network. During the period in which the node E sends the CNM message to the source terminal to reduce the rate of sending data, the rate at which the outbound port at the node E sends data is controlled, but the PFC mechanism has not been started. In the short time, the node E does not control the rate of receiving data at the ingress port, which causes the congestion of the egress port of the node E to be more serious.

In the prior art, the QCN mechanism and the PFC mechanism work together by setting a threshold. However, as described above, once the threshold setting is unreasonable, it is likely to cause congestion spread or congestion, which reduces the efficiency of the entire network system. In addition, due to the complexity and unpredictability of data traffic on the network, the fixed threshold setting method cannot effectively respond to changes in network data traffic, and cannot solve the problem of congestion control flexibly.

Summary of the invention

In view of this, the embodiments of the present invention provide a method, a device, and a system for congestion control, which can effectively cope with changes in network data traffic, thereby flexibly solving the problem of congestion control.

In a first aspect, an embodiment of the present invention provides a method for congestion control, including:

The first network node performs priority-based flow control (PFC) detection on the ingress port, for example, detecting that a certain data stream is congested within the set detection period, and obtaining a flow identifier of the data stream;

Generating, by the first network node, a PFC frame, where the PFC frame includes a flow identifier of the data stream;

The first network node sends the PFC frame to a second network node, instructing the second network node to perform congestion control on the data flow.

In a second aspect, the embodiment of the present invention further provides a method for congestion control, including:

The network node receives a priority-based flow control (PFC) frame, where the PFC frame includes a flow identifier of the data stream;

The network node performs congestion control on the data stream according to the PFC frame.

In a third aspect, an embodiment of the present invention further provides a network node, including:

The detection module is configured to perform priority-based flow control (PFC) detection on the ingress port, such as detecting that a certain data stream is congested within a set detection period, and obtaining a flow of the data stream. Identification

a PFC frame generating module, configured to generate a PFC frame, where the PFC frame includes a stream identifier of the data stream;

The PFC frame sending module is configured to send the PFC frame, where the PFC frame is used to indicate that the network node that receives the PFC frame performs congestion control on the data stream.

In a fourth aspect, the embodiment of the present invention further provides a network node, including:

a receiving module, configured to receive a priority-based flow control (PFC) frame, where the PFC frame includes a flow identifier of a certain data stream;

a congestion control module, configured to perform congestion control on the data flow according to the PFC frame.

In a fifth aspect, an embodiment of the present invention further provides a congestion control system, including a first network node and a second network node;

The first network node is configured to perform priority-based flow control (PFC) detection on the ingress port, such as detecting congestion of a certain data flow in the set detection period, and obtaining the data. a flow identifier of the flow; the first network node is configured to generate a PFC frame that includes a flow identifier of the data flow, and send the PFC frame to the second network node;

The second network node is configured to perform congestion control on the data flow.

In a sixth aspect, an embodiment of the present invention further provides a network node, including a processor and a memory. The memory stores execution instructions that, when the network node is running, communicate with the memory, the processor executing the execution instructions to cause the network node to perform the method of the first aspect above.

In a seventh aspect, an embodiment of the present invention further provides a network node, including a processor and a memory, where the memory stores an execution instruction, and when the network node is running, the processor communicates with the memory, where Executing the execution instructions by the processor causes the network to perform the method of the second aspect above.

In an eighth aspect, an embodiment of the present invention further provides a computer readable medium, comprising computer execution instructions, for causing a host to perform the method of the above first aspect.

In a ninth aspect, the embodiment of the present invention further provides a computer readable medium, comprising computer execution instructions, wherein the computer execution instructions are used to cause a host to perform the method of the second aspect.

The embodiment of the present invention improves the existing PFC mechanism, so that the PFC mechanism can perform congestion detection on a specific data flow, and the congestion degree is sent to the uplink node through the improved PFC, which can be based on the congestion situation. Flexible processing of data streams can effectively cope with changes in network data traffic and congestion and increase the efficiency of the entire network system.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present invention, Those skilled in the art can also obtain other drawings based on these drawings without paying any creative work.

1 is a schematic diagram showing the working principle of a QCN mechanism in the prior art;

2 is a schematic diagram of a working principle of a PFC mechanism in the prior art;

3 is a schematic diagram of a working principle of a common application of a QCN mechanism and a PFC mechanism in the prior art;

4 is a schematic diagram of an improvement of a PFC structure according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a working principle of Embodiment 1 of the present invention; FIG.

FIG. 6 is a schematic flowchart of a method according to Embodiment 1 of the present invention; FIG.

7 is a schematic structural diagram of a network node according to Embodiment 3 of the present invention;

8 is a schematic structural diagram of a network node according to Embodiment 4 of the present invention;

FIG. 9 is a schematic structural diagram of a system according to Embodiment 5 of the present invention.

detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are a part of the embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

As shown in FIG. 4, it is an improvement of the existing PFC structure in the embodiment of the present invention. The existing PFC桢 includes the following fields:

Destination address: 6 bytes, used to indicate the destination MAC address of the data port.

Source address: 6 bytes, used to indicate the source MAC address of the data port.

Ethetype: 2 bytes, used to indicate the type of 桢.

Control opcode: 2 bytes, used to indicate the control code.

Prioirty enable vector: 2 bytes, expressed as backpressure enable vector, where the first byte is directly set to 0; the second byte of E(N) is 8 bits respectively defined in the PFC mechanism. Corresponding to the level queue, indicating whether it is necessary to suspend the transmission of data on the priority queue N; when E(N)=1, it indicates that the data on the priority queue N needs to be suspended, and the pause time is the next Time(N); When E(N) = 0, it means that it is not necessary to suspend the transmission of data on the priority queue N.

Time(0)-Time(7): each 2 bytes, which is a pause timer, indicating the duration of data transmission for suspending the corresponding priority queue. The unit can be the time used to transmit 512-bit data at the transmission rate of the current physical chip. For example, E(3)=1 and Time(3)=8 in the Priority enable vector, the PFC frame indicates the data transmission on the pause priority queue 3, and the pause duration is sent at the transmission rate of the current physical chip. The time required for 512=4096 bits of data.

Pad: Reserved field, 26 bytes.

CRC: 4 bytes, which is cyclic redundancy check information.

In the embodiment of the present invention, the following information may be added in the reserved Pad field:

Flow Identifier (Flow ID): 2 bytes, which is the flow identifier. The Flow ID is the identifier assigned by the terminal that supports the QCN mechanism. The flow ID can be used to find the specific data flow caused by multiple data streams of the source terminal. For the PFC mechanism, when detecting that the ingress port is congested, the network node may mark the data flow that causes congestion;

Times: 2bit, indicating that the number of times the data stream corresponding to the Flow ID is congested is detected within a set period;

Label: 1 bit, marking the current flow control state. The value can be Red or Green. According to the actual design, Red can take a value of 0 or 1, correspondingly Green is 1 or 0. If Label=Red, the indication is The node that receives the PFC flow control PFC桢 triggers the QCN flow control mechanism; if Label=Green, the QCN flow control mechanism is not triggered. Label and Times are associated. Generally, N_Red is set on the node to determine the flow control state. If the value of Times exceeds N_Red, Label=Green, otherwise Label=Red.

FIG. 5 is a schematic diagram of the working principle of the first embodiment of the present invention. In this embodiment, the QCN mechanism and the PFC mechanism are applied together, the network node E supports the QCN mechanism and the PFC mechanism; A and D, B and D, D and E, C and E, and F and E use PFC mechanism for flow control . A threshold for triggering the PFC mechanism and a threshold for triggering the QCN mechanism may be set on each node, and the network node determines whether congestion occurs and the severity of congestion according to the cache usage on the node and the foregoing threshold. Taking the data flow Flow1 (Flow ID=FID1) sent from the source terminal to the destination terminal as shown in FIG. 5 as an example, the data stream flows through the nodes A to D and flows from the node E to the destination terminal. The working principle of the method in the first embodiment of the present invention is described in detail below with reference to the flow diagram 6.

S601: The PFC mechanism on the node E starts to work, that is, the node E performs PFC detection on the ingress port. When the node E detects that the ingress port is congested, it marks the flow ID of the data flow Flow1 that caused the congestion, that is, FID1, and generates a PFC as shown in FIG.

If congestion is not detected in the ingress port within the set period T, the node E continues to perform PFC detection on the ingress port.

S603: If congestion occurs on Flow1 in the set period T, and the number n of congestions of Flow1 is detected, node E generates PFC桢, and sends the generated PFC桢 to the node along the uplink direction of the data flow Flow1. The upstream node of E is node D. The value of the Flow ID field in the PFC is FID1, and the value in the Times field is n. If n is smaller than the pre-set N_Red, the value of the Label field is Green. If n is greater than N_Red, the value of the Label field is Red.

S605: When Label=Green in the PFC桢, the node D pauses to send the data on the Flow1 to the node E within the set time, and performs the PFC detection on the ingress port of the flow1. The time of this setting is also called the pause duration (ie Pause Duration).

S607: If the congestion of the Flow1 is detected on the node D during the pause period, the node D directly triggers the QCN flow control mechanism, that is, the node D generates a CNM message and sends the message to the source terminal of the data flow Flow1, informing the source terminal to reduce the corresponding data flow. Flow1 data transmission rate.

S609: If the node E detects that the number of congestions on Flow1 exceeds N_Red in the period T, that is, Label=Red, after receiving the PFC, the node D pauses to send the data on the Flow1 to the node E, generates a CNM message, and sends the message to the data stream. The source terminal of Flow1 notifies the source terminal to reduce the transmission rate of the data stream Flow1.

S611: After the pause period ends, the data on the data stream Flow1 is sent back to the node E on the node D.

Specifically, the PFC mechanism on the node D does not detect that Flow1 is congested during the pause period, and the node D restores the data on the data stream Flow1 to the node E. In the case of the above S607 or S609, after the pause period ends, the node D also resumes transmitting the data on the data stream Flow1 to the node E; of course, the data transmission on the Flow1 can be resumed at the same time as or before the node D sends the CNM message. As long as it is after the end of the suspension period.

At the same time as the data transmission is resumed, both node D and node E will clear the previously marked Label and Times parameter values on the Flow1.

It should be noted that, in the above embodiment of the present invention, the node D pauses to send the data on the Flow1 to the node E during the set pause period, and adopts the pause mechanism in the PFC mechanism. For details, refer to the improved PFC structure shown in Figure 4. If node E needs node D to suspend the data stream transmission on Flow1, the backpressure enable vector Prioirty enable vector of the PFC frame will be the best for Flow1. The E(N) corresponding to the priority queue N is set to 1, and the pause period Time(N) is set. After receiving the PFC frame, the node E pauses the data transmission of the priority queue N, and the duration is Time (N); therefore, other data streams of the priority queue N are also suspended. After the end of the pause period Time(N), the node D restores the data transmission of the priority queue N, including the data transmission of the above data stream Flow1.

In the first embodiment of the present invention, the congestion of the data stream is detected by using the PFC mechanism on the ingress port of the network node, and corresponding processing is performed according to the severity of the congestion: the number of congestions of a certain data flow does not exceed the set value within a set period. The congestion condition is sent to the uplink network node in the reverse direction of the data stream through the PFC, and the uplink network node pauses to send the data of the data stream during the pause period, thereby slowing local congestion, and when the uplink network node is suspended. When the data stream is detected to be congested again during the period, the source terminal of the data stream is notified by the QCN flow control mechanism to reduce the transmission rate of the data stream.

Of course, if the network node that performs the PFC mechanism detection on the ingress port detects that a certain data stream exceeds the set value, that is, the congestion is already serious, the uplink network node pauses to send the data stream. At the same time of data, the source terminal of the data stream can be notified by the QCN flow control mechanism to reduce the transmission rate of the data stream. In the actual service, when the source terminal receives the CNM message, it also starts a timer for determining whether the congestion condition is alleviated within the set recovery period; if the congestion is relieved, the transmission rate before the data stream is restored. In an actual network environment, multiple network nodes may notify the source terminal to reduce the transmission rate of the same data stream in the above recovery period (as in the first embodiment, if congestion occurs, node A, node D, and node E are both The source terminal is notified to reduce the transmission rate of the data stream Flow1. The source terminal reduces the transmission rate of the data stream when receiving the CNM message for the first time, and the source terminal may not further forward the CNM message received again during the recovery period. The rate is reduced, but the above recovery period is extended.

It should be further noted that, in the foregoing embodiment of the present invention, when the network node that performs the PFC mechanism detection on the ingress port detects that the number of congestions of a certain data stream exceeds a set value within a set period, the network node sends the network node. The PFC frame is sent to the uplink network node, and the uplink network node pauses to transmit the data of the data stream. The uplink network node sends a CNM message to notify the source terminal of the data stream to reduce the data stream transmission rate through the QCN flow control mechanism. In the actual implementation process, the network node that performs the PFC mechanism detection on the ingress port may also send the PFC frame to the uplink network. The network node causes the uplink network node to suspend transmission of data of the data stream; on the other hand, the network node directly sends a CNM message to the source terminal of the data stream to notify the source terminal to reduce the transmission rate of the data stream.

The above-described first embodiment is merely illustrative of an exemplary embodiment of the present invention. In the implementation process, the PFC structure of Fig. 4 can be further improved and applied to the embodiments illustrated in Figs. 5 and 6 above, as well as to all other embodiments of the present invention. For example, the PFC can only include the Flow Identifier and Label fields, and does not include the Times field. It can also include only the Flow Identifier and Times fields, and does not include the Label field. In this case, the node E only needs to record the number of times of congestion on Flow1 in the period T. n, you do not need to set N_Red and compare with n. After receiving PFC桢 on node D, you can use the Times field in the received PFC桢 (that is, the number of times that node E detects congestion on Flow1 in period T). n) and the node D detects the degree of congestion and the degree of congestion on the Flow1 by the PFC mechanism during the pause period to determine the degree of the race of the flow1 and determine the operation of the node D, such as n+m is greater than the set value to trigger the QCN flow. The control mechanism sends a CNM message; or the CNM message may be sent when m is greater than the set value to trigger the QCN flow control mechanism. Of course, the PFC桢 can also include only the Flow Identifier field, and does not include the Label and Times fields. After the node D receives the PFC sent by the node E, the node D only detects whether congestion and congestion occur on the Flow1 through the PFC mechanism during the pause period. The number m is used to determine the degree of the game of the flow1 and determine the operation of the node D. If the value of m is greater than the set value, the QCN flow control mechanism is triggered to send the CNM message, or the number of times the flow1 is congested during the pause period may not be recorded. Send a CNM message.

Embodiment 2 of the present invention provides a method for congestion control, including:

The first network node performs priority-based flow control (PFC) detection on the ingress port, and detects that a certain data flow, such as Flow1, is congested within the set detection period, and obtains the flow identifier of Flow1. Flow ID;

The first network node generates a PFC frame and sends the PFC frame to the second network node, where the flow identifier Flow ID of Flow1 is included;

The second network node performs congestion control on Flow1.

Optionally, the foregoing PFC frame further includes congestion degree information, and the second network node is configured to perform congestion control on the Flow1 according to the congestion degree information. The congestion degree information is a parameter used to indicate the congestion severity of the data flow Flow1. For the structure diagram of the PFC frame shown in FIG. 4, it may be expressed by a Label field or a Times field, or may be expressed by a Label field and a Times field. For example, Label=Red indicates that the congestion is serious, and Label=Green indicates that the congestion is light. The congestion can be first relieved by locally suspending the transmission of the data stream; of course, the number of congestions on Flow1 in the period T can be directly used. Judging the severity of congestion, the larger the value of n, the more serious the congestion.

Optionally, the second network node performing congestion control on the flow1 according to the congestion degree information may include:

When the congestion degree information indicates that the congestion is serious, the second network node sends a congestion notification message (CNM) to the source terminal of the data flow Flow1, and notifies the source terminal to reduce the transmission rate of the data flow Flow1; or

When the congestion degree information indicates that the congestion is light, the second network node pauses to send the data of the data stream Flow1 to the first network node during the set pause period; the second network node may also enter the port pair data stream during the pause period. Flow1 performs PFC detection: If the PFC detects that the data flow Flow1 is congested during the suspension period, the second network node may send a CNM message to the source terminal of the data flow Flow1, and notify the source terminal to reduce the transmission rate of the data flow Flow1.

After the suspension period ends, the second network node resumes transmitting data of the data stream Flow1 to the first network node.

Optionally, the first network node detects that the data flow Flow1 is seriously congested during the detection period (for example, when the number of congestions exceeds a set value or the cache occupancy rate is too high), the CNM message may be directly sent to the source of the data flow Flow1. The terminal notifies the source terminal to reduce the transmission rate of the data stream.

The first embodiment and the second embodiment of the present invention improve the existing PFC mechanism and the PFC structure, so that when the network node performs the PFC mechanism detection, it can determine congestion detection of a specific data stream that is congested, and The congestion degree information is sent to the uplink node through the improved PFC. When the congestion is not serious, the transmission rate of the data stream is locally reduced first to alleviate the congestion; when the data stream still has congestion after the congestion is severe or the local congestion is congested, the source terminal of the data flow is notified by the QCN flow control mechanism to reduce the data. The rate at which the stream is sent. By using the method in the first embodiment of the present invention, the data stream can be flexibly processed according to the congestion condition, which can effectively cope with the problem of network data traffic change and congestion spread, and improve the efficiency of the entire network system.

Embodiment 3 of the present invention provides a network node 700, as shown in FIG. The network node 700 includes:

The detecting module 701 is configured to perform priority-based flow control (PFC) detection on the ingress port, such as detecting that a certain data stream, such as Flow1, is congested within the set detection period, and obtaining the data stream. Flow ID of Flow1;

The PFC frame generating module 703 is configured to generate a PFC frame, where the flow identifier of the data stream Flow1 obtained by the detecting module 701 is included;

The PFC frame sending module 705 is configured to send a PFC frame generated by the PFC frame generating module 703, where the PFC frame is used to indicate that the network node that receives the PFC frame performs congestion control on the data stream Flow1.

Optionally, the foregoing PFC frame further includes congestion degree information. The congestion degree information is a parameter indicating the congestion severity of the data flow Flow1. For the structure diagram of the PFC frame shown in FIG. 4, it may be expressed by a Label field or a Times field, or may be expressed by a Label field and a Times field. For example, Label=Red indicates that the congestion is serious, Label=Green indicates that the congestion is light, and the congestion can be alleviated by locally suspending the transmission of the data stream; of course, the number of times of congestion on Flow1 can be directly used in the set detection period. n to determine the severity of congestion, the greater the value of n, the more serious congestion.

Optionally, the network node may further include a CNM message sending module 707, configured to detect that the data flow Flow1 is severely congested during the detection period (for example, the number of times of congestion exceeds a set value or is slow When the storage occupancy rate is too high, the source terminal that sends the CNM message to the data stream notifies the source terminal to reduce the transmission rate of the data stream.

The network node provided in the foregoing Embodiment 3 of the present invention can identify which data stream is congested and provide it to other network nodes to perform more flexible congestion control on the data stream; further, the congestion degree information can be sent to The uplink node firstly reduces the transmission rate of the data stream to reduce congestion when the congestion is not serious; when the data stream still appears to be congested after the congestion is severe or the local congestion is congested, the source terminal of the data flow is notified by the QCN mechanism to reduce the data transmission rate. . In this embodiment, the data stream is flexibly processed according to the congestion condition, which can effectively cope with the problem of network data traffic change and congestion spread, and improve the efficiency of the entire network system.

Embodiment 4 of the present invention provides a network node 800, as shown in FIG. The network node 800 includes:

The receiving module 801 is configured to receive a priority-based flow control (PFC) frame, where the PFC frame includes a flow identifier of a certain data flow, such as Flow1;

The congestion control module 803 is configured to perform congestion control on the data flow Flow1 according to the PFC frame received by the receiving module 801.

Optionally, the PFC frame may further include congestion degree information; the congestion control module 803 performs congestion control on the data flow Flow1 according to the congestion degree information. The congestion degree information is a parameter indicating the congestion severity of the data flow Flow1. For the structure diagram of the PFC frame shown in FIG. 4, it may be expressed by a Label field or a Times field, or may be expressed by a Label field and a Times field. For example, Label=Red indicates that the congestion is serious, Label=Green indicates that the congestion is light, and the congestion can be first relieved by locally suspending the transmission of the data stream; of course, the number of times of congestion on the data flow Flow1 during the detection period can be directly used. To determine the severity of congestion, the larger the value of n, the more serious the congestion.

The optional congestion control module 803 can also be used to suspend the transmission during the set pause period. The data of the data stream.

Optionally, the network node 800 may further include: a sending module 805. The congestion control module 803 is further configured to perform PFC detection on the data stream Flow1 at the ingress port during the pause period; if the data stream Flow1 is detected to be congested during the pause period, the sending module 805 is configured to send a congestion notification message (Congestion Notification Message, CNM) to the source terminal of the data stream Flow1, notifying the source terminal to reduce the transmission rate of the data stream Flow1.

After the pause period ends, the congestion control module 803 resumes transmitting the data of the data stream Flow1.

The network node provided in the foregoing Embodiment 4 of the present invention can identify which data stream is congested according to the received PFC frame, and perform more flexible congestion control on the data flow according to the degree of congestion: firstly when the congestion is not serious The transmission rate of the data stream is reduced to alleviate the congestion; when the data stream still has congestion after the congestion is severe or the local congestion is congested, the source terminal of the data stream is notified by the QCN mechanism to reduce the data transmission rate. In this embodiment, the data stream is flexibly processed according to the congestion condition, which can effectively cope with the problem of network data traffic change and congestion spread, and improve the efficiency of the entire network system.

Embodiment 5 of the present invention provides a congestion control system. As shown in FIG. 9, the system includes a first network node and a second network node.

The first network node performs priority-based flow control (PFC) detection on the ingress port, and if a certain data stream, such as Flow1, is detected to be congested within a set detection period, the data stream is acquired. Flow ID of Flow1. The first network node generates a PFC frame including the flow identifier of the data stream Flow1, and transmits the PFC frame to the second network node.

The second network node performs congestion control on the data flow Flow1 according to the PFC.

Optionally, the PFC frame may further include congestion degree information; and the second network node performs congestion control on the data flow Flow1 according to the congestion degree information. The congestion degree information is a parameter used to indicate the congestion severity of the data flow Flow1. See the structure diagram of the PFC frame shown in FIG. Expressed through the Label field or the Times field, it can also be expressed by the Label field and the Times field. For example, Label=Red indicates that the congestion is serious, Label=Green indicates that the congestion is light, and the congestion can be first relieved by locally suspending the transmission of the data stream; of course, it can also be directly judged by the number n of congestions on Flow1 in the detection period. The severity of congestion, the greater the value of n, the more severe the congestion.

Optionally, the second network node according to the congestion control on the data flow Flow1 may include:

The second network node is configured to send a Congestion Notification Message (CNM) to the source terminal of the data flow Flow1, and notify the source terminal to reduce the transmission rate of the data flow Flow1.

Optionally, the second network node according to the congestion control of the data flow Flow1 may further include: the second network node suspending sending the data of the data flow Flow1 to the first network node during the set pause period.

Further, the second network node may further perform PFC detection on the data flow Flow1 at the ingress port during the pause period;

If the data flow Flow1 is congested during the suspension period, the second network node sends a CNM message to the source terminal of the data flow Flow1, informing the source terminal to decrease the transmission rate of the data flow.

Optionally, the first network node may directly send the CNM message to the data stream when detecting that a serious congestion occurs in the set detection period (for example, when the number of congestion of the data flow Flow1 exceeds a set value or the cache occupancy rate is too high) The source terminal notifies the source terminal to reduce the transmission rate of the data stream.

The congestion control system provided in the above fifth embodiment of the present invention can identify which data stream is sent. Congestion occurs and more flexible congestion control is performed according to the degree of congestion: when the congestion is not serious, the transmission rate of the data stream is locally reduced to alleviate the congestion; when the data stream still has congestion after the congestion is severe or the local congestion is congested, the notification is notified through the QCN mechanism. The source terminal of the data stream reduces the transmission rate of the data stream. In this embodiment, the data stream is flexibly processed according to the congestion condition, which can effectively cope with the problem of network data traffic change and congestion spread, and improve the efficiency of the entire network system.

Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented in hardware, firmware implementation, or a combination thereof. When implemented in software, the functions described above may be stored in or transmitted as one or more instructions or code on a computer readable medium. Computer readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one location to another. A storage medium may be any available media that can be accessed by a computer. By way of example and not limitation, computer readable media may comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage media or other magnetic storage device, or can be used for carrying or storing in the form of an instruction or data structure. The desired program code and any other medium that can be accessed by the computer. Also. Any connection may suitably be a computer readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable , fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, wireless, and microwave are included in the fixing of the associated media. As used in the present invention, a disk and a disc include a compact disc (CD), a laser disc, a compact disc, a digital versatile disc (DVD), a floppy disk, and a Blu-ray disc, wherein the disc is usually magnetically copied, and the disc is The laser is used to optically replicate the data. Combinations of the above should also be included within the scope of the computer readable media.

In summary, the above description is only a preferred embodiment of the technical solution of the present invention, and is not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.

Claims

A method for congestion control, comprising:

The first network node performs priority-based flow control (PFC) detection on the ingress port, for example, detecting that a certain data stream is congested within the set detection period, and obtaining a flow identifier of the data stream;

Generating, by the first network node, a PFC frame, where the PFC frame includes a flow identifier of the data stream;

The first network node sends the PFC frame to a second network node, instructing the second network node to perform congestion control on the data flow.
The method of claim 1, wherein the PFC frame further comprises congestion level information; and the second network node is configured to perform congestion control on the data stream according to the congestion degree information.
The method according to claim 2, wherein the second network node is configured to perform congestion control on the data flow according to the congestion degree information, including:

The second network node is configured to send a congestion notification message (CNM) to a source terminal of the data stream, and notify the source terminal to decrease a transmission rate of the data stream.
The method according to claim 2, wherein the second network node is configured to perform congestion control on the data flow according to the congestion degree information, including:

The second network node is configured to suspend transmission of data of the data stream to the first network node during a set pause period.
The method according to claim 4, wherein the second network node is configured to perform congestion control on the data stream according to the congestion degree information, further comprising:

The second network node is configured to perform PFC detection on the data flow at an ingress port during the pause period;

If the data flow is congested during the suspension period, the second network node is configured to send a congestion notification message (CNM) to the source terminal of the data flow, and notify the source terminal to decrease the data flow. The sending rate.
The method according to claim 1, 2 or 4, wherein the first network node sends a congestion notification message (CNM) when the data stream is detected to be heavily congested during the detection period. And to the source terminal of the data stream, informing the source terminal to decrease the transmission rate of the data stream.
A method for congestion control, comprising:

The network node receives a priority-based flow control (PFC) frame, where the PFC frame includes a flow identifier of the data stream;

The network node performs congestion control on the data stream according to the PFC frame.
The method according to claim 7, wherein said PFC frame further comprises congestion degree information; said network node performs congestion control on said data stream according to said congestion degree information.
The method according to claim 8, wherein the network node performs congestion control on the data stream according to the congestion degree information, including:

The network node sends a congestion notification message (CNM) to the source terminal of the data stream, and notifies the source terminal to decrease the transmission rate of the data stream.
The method according to claim 7, wherein the network node performs congestion control on the data stream according to the congestion degree information, including:

The network node suspends transmitting data of the data stream during a set pause period.
The method of claim 10, wherein the controlling, by the network node, the congestion control of the data stream according to the congestion degree information further comprises:

The network node performs PFC detection on the data stream at the ingress port during the pause period;

If the data stream is congested during the suspension period, the network node sends a Congestion Notification Message (CNM) to the source terminal of the data stream, and notifies the source terminal to decrease the transmission rate of the data stream.
A network node, comprising:

The detection module is configured to perform priority-based flow control (PFC) detection on the ingress port, such as detecting that a certain data stream is congested within a set detection period, and obtaining a flow of the data stream. Identification

a PFC frame generating module, configured to generate a PFC frame, where the PFC frame includes a stream identifier of the data stream;

The PFC frame sending module is configured to send the PFC frame, where the PFC frame is used to indicate that the network node that receives the PFC frame performs congestion control on the data stream.
The network node of claim 12 wherein said PFC frame further comprises congestion level information.
The network node according to claim 12 or 13, wherein the network node further comprises a CNM message sending module, configured to send a CNM message to the device when detecting that the data stream is seriously congested within the detecting period The source terminal of the data stream notifies the source terminal to decrease the transmission rate of the data stream.
A network node, comprising:

a receiving module, configured to receive a priority-based flow control (PFC) frame, where the PFC frame includes a flow identifier of a certain data stream;

a congestion control module, configured to perform congestion control on the data flow according to the PFC frame.
The network node according to claim 15, wherein the PFC frame further includes congestion degree information; and the congestion control module performs congestion control on the data flow according to the congestion degree information.
The network node according to claim 16, further comprising:

The sending module is configured to send a congestion notification message (CNM) to the source terminal of the data stream, and notify the source terminal to decrease the sending rate of the data stream.
The network node of claim 15 wherein said congestion control module is operative to suspend transmission of data of said data stream during a set pause period.
The network node according to claim 18, wherein said congestion control module is configured to perform PFC detection on said data stream at an ingress port during said pause period;

If the data flow is congested during the suspension period, the congestion control module sends a congestion notification message (CNM) to the source terminal of the data flow, and the source terminal is notified to reduce the transmission rate of the data flow. .
A congestion control system, comprising: a first network node and a second network node;

The first network node is configured to perform priority-based flow control (PFC) detection on the ingress port, such as detecting in a set detection period. Congesting to a certain data stream, obtaining a flow identifier of the data stream; the first network node is configured to generate a PFC frame that includes a flow identifier of the data flow, and send the PFC frame to the second network node ;

The second network node is configured to perform congestion control on the data flow.
The system of claim 20, wherein the PFC frame further comprises congestion level information; and the second network node is configured to perform congestion control on the data stream according to the congestion degree information.
The system according to claim 21, wherein the second network node is configured to perform congestion control on the data flow according to the congestion degree information, including:

The second network node is configured to send a congestion notification message (CNM) to a source terminal of the data stream, and notify the source terminal to decrease a transmission rate of the data stream.
The system according to claim 21, wherein the second network node is configured to perform congestion control on the data flow according to the congestion degree information, including:

The second network node is configured to suspend transmission of data of the data stream to the first network node during a set pause period.
The system according to claim 23, wherein the second network node is configured to perform congestion control on the data stream according to the congestion degree information, further comprising:

The second network node is configured to perform PFC detection on the data flow at an ingress port during the pause period;

If the data stream is congested during the suspension period, the second network node is configured to send a Congestion Notification Message (CNM) to the source end of the data stream. End, notifying the source terminal to reduce the sending rate of the data stream.
The system of claim 20, 21 or 23, wherein the first network node is further configured to send a Congestion Notification Message, when the data stream is detected to be heavily congested during a detection period. CNM) to the source terminal of the data stream, notifying the source terminal to reduce the transmission rate of the data stream.
A network node, comprising: a processor and a memory, the memory storing execution instructions, the processor communicating with the memory when the network node is running, the processor performing the execution The instructions cause the network to perform the method of any of claims 1 to 6.
A network node, comprising: a processor and a memory, the memory storing execution instructions, the processor communicating with the memory when the network node is running, the processor performing the execution The instructions cause the network to perform the method of any of claims 7-11.
A computer readable medium comprising computer executed instructions for causing a host to perform the method of any one of claims 1 to 6.
A computer readable medium comprising computer executed instructions for causing a host to perform the method of any one of claims 7 to 11.