A switch cache management method for distinguishing monitoring and exchange data

Abstract

The present invention relates to a switch cache management method for distinguishing between monitoring and exchange data, and belongs to the field of computer communications. The method comprises: providing a shared cache at the input end of the switch for storing exchange data and monitoring data corresponding to all input ports; providing a queue at each output port of the switch for storing management information of the exchange data and monitoring data corresponding to the corresponding output port; based on the shared cache and the queue, each output port adopts a different management mechanism for sending the corresponding exchange data and monitoring data; a first-in-first-out sequential sending mechanism is adopted for monitoring data, and a fiber optic network communication control method with a feedback mechanism is used to control the sending of exchange data. This method increases the controllability of network status, avoids network congestion for exchange data, and prevents the impact of monitoring services on the normal communication service performance of the switch, effectively improving the cache utilization rate of the switch and enhancing switch performance.

Description

Technical Field

[0001] The invention belongs to the field of computer communications, and in particular relates to a switch buffer management method for distinguishing between monitoring and exchanging data. Background Art

[0002] With the continuous development of aerospace electronics systems, the demand for switching in switching network systems is also growing. In traditional switch architectures, increasing bandwidth, data throughput, and transmission rates make it increasingly difficult to monitor and manage port data flows. This leads to a series of common problems in actual switch applications, such as increased difficulty in real-time online analysis, longer fault analysis and location cycles, and increasingly low troubleshooting efficiency.

[0003] The ports of mainstream switches primarily perform two functions: switching and monitoring. Switching enables basic data exchange between two ports, while monitoring uses a specific port to monitor the input and output data of any port. Input or output data frames from the monitored port are copied to the monitoring port and output, enabling real-time online analysis of switch data flows and fault monitoring. Currently, switches typically use the same store-and-forward method for both switching data and monitoring data on their output ports, without differentiated scheduling. This causes monitoring services to impact the normal communication performance of the switch, impacting switch performance. Summary of the Invention

[0004] In light of the above analysis, the present invention aims to provide a switch cache management method that distinguishes between monitoring and exchange data. Based on a crossbar structure, exchange data and monitoring data are stored in a shared cache. These data are then scheduled and sent using different forwarding mechanisms. This increases the controllability of network status, improves the communication efficiency of exchanged data, avoids the impact of monitoring services on the normal communication performance of the switch, effectively improves the cache utilization of the switch, and enhances switch performance. Specifically, the method includes the following steps:

[0005] A shared cache is set at the input end of the switch to store the switching data and monitoring data corresponding to all input ports;

[0006] A queue is set at each output port of the switch output end for storing management information of the switching data and monitoring data corresponding to the corresponding output port;

[0007] Based on the shared buffer and the queue, each output port adopts different management mechanisms to send corresponding exchange data and monitoring data.

[0008] Furthermore, the step of setting a shared cache at the input end of the switch for storing the exchange data and the monitoring data includes:

[0009] Based on the crossbar architecture, a shared cache RAM is set at the input end of the switch to store the switching data and monitoring data corresponding to all input ports.

[0010] Furthermore, setting a queue at each output port of the switch output end for storing management information of exchanged data and monitored data includes setting a switching management queue and a monitoring management queue at each output port of the switch, which are respectively used to store management information of exchanged data and management information of monitored data corresponding to the output port.

[0011] Furthermore, the sending of exchange data and monitoring data by adopting different management mechanisms includes:

[0012] The target output port of the monitored data sends the monitored data in the first-in-first-out order at this port;

[0013] The target output port of the exchanged data uses the optical fiber network communication control method with a feedback mechanism to send the exchanged data at the port.

[0014] Furthermore, the optical fiber network communication control method with a feedback mechanism for sending exchange data includes:

[0015] The payload priority of the frame is determined based on the payload size; the application priority of the frame is determined based on the priority specified by the application layer; wherein the frame refers to the exchanged data.

[0016] Authorizing frames to be sent based on the number of payload priority levels, the payload priority level, and the application priority level of the frames to be sent;

[0017] The link congestion is determined based on the average time interval of the local receiving primitive R_RDY and the upper limit of the reply R_RDY time;

[0018] Based on the link congestion situation and local credit value feedback, the link communication is controlled to determine whether to send the authorized frames to be sent.

[0019] Furthermore, determining the load priority of the frame based on the load size includes:

[0020] Determine the priority of a frame with a payload byte number less than or equal to l1 bytes as a first payload priority; where l1 is a preset value, 1<l1<32;

[0021] Determine the priority of a frame with a payload byte number greater than l1 bytes and less than or equal to l2 bytes as a second payload priority, where l2 is a preset value, 480<l2<1024;

[0022] The priority of the frame whose payload byte number is greater than 12 bytes is determined to be the third payload priority.

[0023] Furthermore, determining the application priority of the frame based on the priority specified by the application layer includes pre-setting different application priorities for different frames according to actual application situations.

[0024] Furthermore, the authorizing the frame to be sent based on the number of load priority levels of the frame to be sent, the level of the load priority, and the level of the application priority includes:

[0025] When only the first load priority frame sends a request, authorizing the first load priority frame;

[0026] When there are frames with more than two load priority levels that have transmission requests, the load priority level for authorizing the frame to be sent is determined based on the historical number of transmissions and the current local credit value, and the frames to be sent at this level are authorized.

[0027] Furthermore, judging the link congestion based on the average time interval of the local reception primitive R_RDY and the upper limit of the reply R_RDY time includes:

[0028] Calculate the latest R_RDY interval of receiving primitives;

[0029] Determine link congestion: If the latest interval is less than a times the average value, the link is considered to be free of congestion and communication is normal; a is a preset value. If it is greater than or equal to a times the average value and less than the upper limit, the link is considered congested and is determined to be congestion level 1. If it is greater than or equal to the upper limit, the link is considered to be extremely congested and is determined to be congestion level 2.

[0030] Furthermore, controlling link communication based on link congestion and local credit value feedback to determine whether to send an authorized frame to be sent includes:

[0031] When the local credit value and link congestion do not support frame transmission and credit usage, skip this transmission and retain the status for the next transmission decision; where the situations where frame transmission and credit usage are not supported include: the local credit value is full; the local credit value is not full but greater than or equal to x1% of the threshold, and the link is congested 1 degree or congested 2 degrees; the local credit value is greater than or equal to x2% of the threshold and less than x1% of the threshold, and the congestion is 2 degrees; where x1 and x2 are preset values, 70 <x1≤99,50<x2≤70;

[0032] When the local credit value is greater than or equal to x3% of the threshold: if there is an authorized frame of the first load priority, it is determined to send the authorized frame of the first load priority until the local credit value is full; if there is no authorized frame of the first load priority, skip this transmission and retain the state for the next transmission decision; x3 is a preset value; 70 <x3≤99;

[0033] When the local credit value is less than x3% of the threshold: frames of all authorized levels are allowed to be sent.

[0034] The present invention can achieve at least one of the following beneficial effects:

[0035] By setting up a shared cache, the exchange data and monitoring data at the input end are stored separately, thereby achieving logical isolation of the exchange data and monitoring data; by setting up a switching management queue and a monitoring management queue, the management information of the exchange data and the management information of the monitoring data are stored separately, and different forwarding mechanisms are adopted for the exchange data and the monitoring data, thereby avoiding the impact of the monitoring service on the normal communication service performance of the switch and improving the performance of the switch.

[0036] By using an optical fiber network communication control method with a feedback mechanism for exchanged data, the forwarding order of exchanged data is determined, the congestion of the link is fed back based on the time of the local receiving primitive R_RDY, and traffic management is performed based on the credit value, thereby increasing the controllability of the network status, avoiding network congestion for exchanged data, and effectively increasing bandwidth utilization while ensuring smooth communication, thereby greatly improving communication efficiency; the frame sending priority is determined by the payload size of the exchanged data frame and the priority specified by the application layer, and the frame sending order is adjusted according to the bandwidth allocation and priority, thereby ensuring the real-time transmission and the efficient operation of the entire network.

[0037] By storing and forwarding switching data and monitoring data separately, when a switch fails, the monitoring data can be quickly located, improving the efficiency of switch fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like parts throughout the drawings.

[0039] Figure 1 This is a schematic diagram of a switch cache management mechanism according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0041] Example 1

[0042] A specific embodiment of the present invention discloses a switch cache management method for distinguishing between monitoring and exchanging data. The switch of this embodiment may be a switch with a Crossbar architecture.

[0043] This embodiment specifically includes the following steps:

[0044] S01. A shared cache is set at the input end of the switch to store exchange data and monitoring data corresponding to all input ports, and a queue is set at each output port of the switch to store management information of the exchange data and monitoring data of the corresponding output port.

[0045] Specifically, such as Figure 1 As shown, based on the crossbar architecture, a shared cache RAM is set at the input end of the switch for all input ports to store the exchange data and monitoring data corresponding to all input ports. The monitoring data refers to the monitoring data obtained by the monitoring function of the switch after copying the exchange data.

[0046] Optionally, the size of the shared buffer is determined based on the credit value of each input port, the maximum value of the single-pin load, and the number of ports.

[0047] Preferably, when the credit value of each input port of the switch is the same, the size of the shared buffer is set to: not less than (credit value of the input port * maximum single-frame load * number of ports) -; wherein the single frame refers to a single piece of exchanged data or monitored data.

[0048] Optionally, when the credit value of each input port of the switch is different, the size of the shared buffer is set to: ∑ the credit value of each port * the maximum single-pin load.

[0049] Specifically, a switching management queue and a monitoring management queue are set at each output port of the switch, respectively used to store management information of switching data and management information of monitoring data corresponding to the output port.

[0050] Specifically, the management information of exchanged data includes RAM storage address information of exchanged data, target output port information of exchanged data and priority of exchanged data; the management information of monitored data includes RAM storage address information of monitored data and target output port information of monitored data.

[0051] Optionally, each output port of the switch can be set as a separate monitoring port, a separate switching interface or a switching and monitoring mixed port. At each output port, a queue is set to store management information of the switching data and management information of the monitoring data respectively.

[0052] Specifically, each queue has equal right to use the shared buffer.

[0053] S02: Based on the shared cache and the queue, each output port adopts different management mechanisms to send corresponding exchange data and monitoring data.

[0054] Specifically, the switching data entering the switch input port is stored in the shared cache and the corresponding management information is stored in the switching management queue corresponding to the target output port; at the same time, a copy of the switching data is converted into monitoring data and stored in the shared cache, and the corresponding management information is stored in the monitoring management queue corresponding to the target output port of the monitoring data.

[0055] Specifically, storing the monitored data into a shared cache and storing its corresponding management information into the monitoring management queue of the target output port of the monitored data includes: determining the target output port based on a pre-set setting, storing the RAM storage address information and the target output port information of the monitored data into the monitoring management queue corresponding to the target output port; wherein the monitoring management queue is a FIFO queue.

[0056] Preferably, the target output port of the monitoring data is a dedicated monitoring port, and each monitoring port distributes the monitoring data in a first-in-first-out manner. Based on the order of the monitoring management queue of the monitoring port, the corresponding RAM storage address of the monitoring data is found to obtain the monitoring data and send it sequentially.

[0057] Specifically, storing the exchange data entering the switch input port into the shared cache and storing the corresponding management information into the exchange management queue corresponding to the target output port includes: searching the routing table to determine the target output port of the exchange data, and storing the RAM storage address information, target output port information and priority of the exchange data into the exchange management queue corresponding to the target output port; wherein the exchange management queue is a FIFO queue.

[0058] Preferably, the target output port for the exchanged data is a separate switch port. Optionally, each target output port uses a fiber optic network communication control method with a feedback mechanism to determine the forwarding order of the exchanged data, obtains the exchanged data based on the RAM storage address of the corresponding exchanged data in the exchange management queue, and transmits it, thereby ensuring normal high-speed operation of communication to the greatest extent possible.

[0059] Specifically, the optical fiber network communication control method with a feedback mechanism is used to determine the forwarding order of exchanged data, including:

[0060] The payload priority of the frame is determined based on the payload size; the application priority of the frame is determined based on the priority specified by the application layer; wherein the frame refers to the exchanged data.

[0061] Authorizing frames to be sent based on the number of payload priority levels, the payload priority level, and the application priority level of the frames to be sent;

[0062] The link congestion is determined based on the average time interval of the local receiving primitive R_RDY and the upper limit of the reply R_RDY time;

[0063] Based on the link congestion situation and local credit value feedback, the link communication is controlled to determine whether to send the authorized frames to be sent.

[0064] Specifically, determining the payload priority of a frame based on the payload size; determining the application priority of a frame based on the priority specified by the application layer includes:

[0065] Specifically, determining the load priority of the frame based on the load size includes: the load priority is negatively correlated with the number of load bytes, and the smaller the number of load bytes, the higher the load priority.

[0066] Specifically, short frames with small loads have the highest priority, which is the first load priority. Preferably, frames with a load byte number less than or equal to l1 bytes have the first load priority; l1 is a preset value, 1<l1<32, and optionally, a register is used to store the l1 value, which is modified according to demand; preferably, l1=16. Such frames are general instruction frames in aerospace systems, and have fast processing response time, which can be processed quickly to reduce bandwidth and cache occupancy.

[0067] Frames with medium payload have the second highest priority, and frames with payload bytes greater than 11 bytes and less than or equal to 12 bytes have the second highest payload priority; where 12 is a preset value, 480 < 12 < 1024; optionally, a register is used to store the value of 12; optionally, 12 = 1024;

[0068] Frames with large payloads have the lowest priority. Optionally, frames with a payload byte count greater than 12 bytes have the third payload priority.

[0069] Specifically, determining the application priority of the frame based on the priority specified by the application layer includes: the priority specified by the application layer refers to pre-setting different application priorities for different frames according to actual application conditions; optionally, it can be set to high, medium, low, or no application priority.

[0070] Specifically, the authorization of the frame to be sent based on the number of payload priority levels, the level of payload priority, and the level of application priority of the frame to be sent includes:

[0071] When only the first load priority frame is requested to be sent, authorizing the first load priority frame to be sent;

[0072] When there are frames with two or more load priority levels that have transmission requests, the load priority level for the frame authorized to be sent is determined based on the historical number of transmissions and the current local credit value, and the frame to be sent is authorized, including:

[0073] Principle 1: When the local credit value is greater than or equal to w1% of the threshold of the local credit value, only authorize the transmission of the first load priority frame; if there is no first load priority frame, do not authorize the frame to be transmitted; where w1 is a preset value, 70 < w1 ≤ 90; optionally, use a register to store the w1 value and modify it according to requirements.

[0074] Principle 2: When the local credit value is greater than or equal to w2% of the threshold and less than w1% of the threshold, if the first load priority frame has not been continuously transmitted in the most recent i times before this transmission, then this time still authorize the first load priority frame to be transmitted, otherwise authorize the second load priority frame to be transmitted; generally, do not authorize the third load priority frame to be transmitted; where i is a preset value, 1 < m < 20; optionally, use a register to store the i value and modify it according to requirements; where w2 is a preset value, 60 < w2 ≤ 70; optionally, use a register to store the w1 value and modify it according to requirements.

[0075] Principle 3: When the local credit value is greater than or equal to w3% of the threshold and less than w2% of the threshold, if the historical transmission times of the first load priority frame are less than or equal to j and the first load priority frame has not been continuously transmitted in the most recent i times before this transmission, then this time authorize the transmission of the first load priority frame to be transmitted, otherwise authorize the second load priority frame or the third load priority frame to be transmitted; where, if and only if there is a third load priority frame waiting to be authorized and the historical transmission times are 0, authorize the third load priority frame to be transmitted, otherwise authorize the second load priority frame to be transmitted; where the historical transmission times refer to the total number of times the load priority frame has been transmitted within a period of time, and the period of time refers to the sum of the transmission processing times of the k transmission frames before the current time; where j and k are preset values, 1 < j, k < 30; optionally, use a register to store the j and k values and modify them according to requirements; w3 is a preset value, 30 < w3 ≤ 60; optionally, use a register to store the w3 value and modify it according to requirements. [[ID=,8]]

[0076] Principle 4: When the local credit value is less than w3% of the threshold, on the basis of authorizing according to the load priority level from high to low, when the following conditions are met, make the load priority frames of each level be transmitted alternately: the first load priority frame cannot be continuously transmitted more than j times, the second load priority frame cannot be continuously transmitted more than t times, and the historical transmission times of the third load priority frame must be no less than s times; where t and s are preset values, 1 < t, s < 15; optionally, use a register to store the t and s values and modify them according to requirements.

[0077] Specifically, in the above allocation principle, when authorizing each level of load priority frame, further sort and authorize based on its application priority; where the order of application priority sorting from high to low is: high application priority, medium application priority, low application priority, no application priority.

[0078] Optionally, the threshold of the above local credit value is a preset constant value; optionally, a register is used to store the threshold of the local credit value and is modified according to requirements.

[0079] Specifically, determining the link congestion situation based on the average value of the time intervals of receiving the local primitive R_RDY and the upper limit value of the reply R_RDY time includes:

[0080] Specifically, R_RDY is a common primitive signal transmitted at the link layer, belonging to the signals specified by the FC protocol and used for flow control at the transport layer; according to the FC_AE protocol, when a local primitive R_RDY is received, a local credit is restored; therefore, the time interval of receiving the primitive can intuitively reflect the current congestion situation of the link.

[0081] Specifically, a local timer is used to calculate the time of receiving the primitive R_RDY; a local list is set up, adopting a first-in-first-out queue structure to record the moments of receiving the primitive R_RDY for every h times of sending a frame and receiving the reply to the message recently and the interval time between this moment and the previous primitive receiving moment; based on all h interval times, the average value is calculated after removing the maximum value and the minimum value; where h is a preset value, 1 < h < 20; optionally, a register is used to store the h value and is modified according to requirements.

[0082] Specifically, the normal network reply R_RDY time should not exceed the upper limit value of the reply R_RDY time.

[0083] Optionally, the upper limit value of the reply R_RDY time is a preset constant value; optionally, a register is used to store the upper limit value of the reply R_RDY time and is modified according to requirements.

[0084] Specifically, determining the link congestion situation based on the average value of the time intervals of receiving the local primitive R_RDY and the upper limit value of the reply R_RDY time includes:

[0085] 1. Calculate the interval time of receiving the current latest primitive R_RDY;

[0086] 2. Judge the link congestion situation:

[0087] If the latest interval time is less than a times the average value, it is considered that there is no congestion in the current link and the communication is normal; where a is a preset value, 1 < a < 5; optionally, a register is used to store the a value and is modified according to requirements.

[0088] If it is greater than or equal to a times the average value and less than the upper limit value, it is considered that the link is congested and determined as congestion degree 1;

[0089] If it is greater than or equal to the upper limit value, it is extremely congested and determined to be congestion level 2.

[0090] Specifically, based on the link congestion degree and local credit value feedback to control link communication, determining whether to send the authorized frames to be sent includes:

[0091] When the local credit value and the link congestion degree do not support frame sending and credit usage, skip this sending and retain this status for the next sending decision; among them, the situations that do not support frame sending and credit usage are: the local credit value is full; the local credit value occupancy is less than full but greater than or equal to x1% of the threshold, and the link is congested at level 1 or congestion level 2; the local credit value occupancy is between greater than or equal to x2% and less than x1% of the threshold, and the congestion level is 2; where x1 and x2 are preset values, 70 < x1 ≤ 99, 50 < x2 ≤ 70; optionally, use registers to store the x1 and x2 values and modify them according to requirements;

[0092] When the local credit value is greater than or equal to x3% of the threshold: If there are frames with the first load priority that have been authorized, the frames with the first load priority that have been authorized can always be sent until the local credit value is full; If there are no frames with the first load priority that have been authorized, skip this sending and retain this status for the next sending decision; where x3 is a preset value, 70 < x3 ≤ 99; optionally, use registers to store the x3 value and modify it according to requirements;

[0093] When the local credit value is less than x3% of the threshold: Allow the sending of authorized frames at each level.

[0094] A switch buffer management method for differentiating listening and exchanging data disclosed in this embodiment separates the exchanged data and the listening data through different storage management and forwarding processes, avoiding the impact of the listening service on the performance of the normal communication service of the switch and improving the performance of the switch.

[0095] In this embodiment, by using a fiber optic network communication control method with a feedback mechanism for the exchanged data to judge the forwarding order of the exchanged data, based on the time of the local receive primitive R_RDY to feedback the congestion situation of the link, and based on the credit value for traffic management, it increases the controllability of the network state, avoids network congestion of the exchanged data, effectively increases the bandwidth utilization rate while ensuring smooth communication, and greatly improves the communication efficiency. Determine the sending priority of the frame according to the load size of the frame of the exchanged data and the priority specified by the application layer, and adjust the sending order of the frame according to the bandwidth allocation and priority to ensure the real-time transmission and the efficient operation of the entire network.

[0096] By storing and forwarding the exchanged data and the listening data separately, when the switch fails, the listening data can be quickly located, improving the fault diagnosis efficiency of the switch.

[0097] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A switch cache management method for distinguishing between monitoring and exchange data, characterized in that: The steps include: A shared cache is set up at the input end of the switch to store the exchange data and monitoring data corresponding to all input ports. The size of the shared cache is determined by the credit value of each input port, the maximum single-frame payload, and the number of ports. Setting a queue at each output port of the switch output end for storing management information of exchanged data and monitored data corresponding to the corresponding output port, including setting a switching management queue and a monitoring management queue at each output port of the switch, respectively for storing management information of exchanged data and management information of monitored data corresponding to the output port; Based on the shared cache and the queue, each of the output ports adopts different management mechanisms to send the corresponding exchange data and monitoring data, including: the target output port of the monitoring data sends the monitoring data in a first-in-first-out order at this port; the target output port of the exchange data uses a fiber optic network communication control method with a feedback mechanism to send exchange data at this port; the fiber optic network communication control method with a feedback mechanism to send exchange data includes: determining the load priority of the frame based on the load size; determining the application priority of the frame based on the priority specified by the application layer; the frame refers to the exchange data; authorizing the frame to be sent based on the number of load priority levels, the level of load priority and the level of application priority of the frame to be sent; judging the link congestion based on the average time interval of the local reception primitive R_RDY and the upper limit of the reply R_RDY time; controlling the link communication based on the link congestion and local credit value feedback to determine whether to send the authorized frame to be sent.

2. The switch cache management method according to claim 1, characterized in that: The step of setting a shared cache at the input end of the switch for storing exchange data and monitoring data includes: Based on the crossbar architecture, a shared cache RAM is set at the input end of the switch to store the switching data and monitoring data corresponding to all input ports.

3. The switch cache management method according to claim 2, characterized in that: Determining the load priority of the frame based on the load size includes: Determine the priority of a frame with a payload byte number less than or equal to l1 bytes as a first payload priority; where l1 is a preset value, 1<l1<32; Determine the priority of a frame with a payload byte number greater than l1 bytes and less than or equal to l2 bytes as a second payload priority, where l2 is a preset value, 480<l2<1024; The priority of the frame whose payload byte number is greater than 12 bytes is determined to be the third payload priority.

4. The switch cache management method according to claim 3, characterized in that: Determining the application priority of the frame based on the priority specified by the application layer includes pre-setting different application priorities for different frames according to actual application situations.

5. The switch cache management method according to claim 3 or 4, characterized in that: The authorizing of the frame to be sent based on the number of load priority levels, the level of load priority and the level of application priority of the frame to be sent includes: When only the first load priority frame sends a request, authorizing the first load priority frame; When there are frames with more than two load priority levels that have transmission requests, the load priority level for authorizing the frame to be sent is determined based on the historical number of transmissions and the current local credit value, and the frames to be sent at this level are authorized.

6. The switch cache management method according to claim 5, characterized in that: The determining of link congestion based on the average value of the time interval for receiving the local primitive R_RDY and the upper limit of the R_RDY reply time includes: Calculate the latest R_RDY interval of receiving primitives; Determine link congestion: If the latest interval is less than a times the average value, the link is considered to be free of congestion and communication is normal; a is a preset value. If it is greater than or equal to a times the average value and less than the upper limit, the link is considered congested and is determined to be congestion level 1. If it is greater than or equal to the upper limit, the link is considered to be extremely congested and is determined to be congestion level 2.

7. The switch cache management method according to claim 6, characterized in that: The controlling link communication based on the link congestion situation and the local credit value feedback and determining whether to send the authorized to-be-sent frame includes: When the local credit value and link congestion do not support frame transmission and credit usage, skip this transmission and retain the status for the next transmission decision; where the situations where frame transmission and credit usage are not supported include: the local credit value is full; the local credit value is not full but greater than or equal to x1% of the threshold, and the link is congested 1 degree or congested 2 degrees; the local credit value is greater than or equal to x2% of the threshold and less than x1% of the threshold, and the congestion is 2 degrees; where x1 and x2 are preset values, 70 <x1≤99,50<x2≤70; When the local credit value is greater than or equal to x3% of the threshold: if there is an authorized frame of the first load priority, it is determined to send the authorized frame of the first load priority until the local credit value is full; if there is no authorized frame of the first load priority, skip this transmission and retain the state for the next transmission decision; x3 is a preset value; 70 <x3≤99; When the local credit value is less than x3% of the threshold: frames of all authorized levels are allowed to be sent.

Images