Congestion control method, apparatus, and system, and computer storage medium
A network device inputs first network status information of the network device in a first time period to an ECN inference model, to obtain an inference result that is output by the ECN inference model based on the first network status information. Then, the network device sends an ECN parameter sample to an analysis device that manages the network device, where the ECN parameter sample includes the first network status information and a target ECN configuration parameter corresponding to the first network status information, and the target ECN configuration parameter is obtained based on the inference result. The network device receives an updated ECN inference model sent by the analysis device.
1. A method implemented by a network device and comprising:
inputting first network status information of the network device in a first time period into an explicit congestion notification (ECN) inference model;
obtaining, based on the first network status information, an inference result from the ECN inference model, wherein the inference result comprises an original ECN configuration parameter;
sending, to an analysis device that manages the network device, an ECN parameter sample comprising a target ECN configuration parameter, wherein the target ECN configuration parameter is based on the inference result; and
receiving, from the analysis device and based on training with the ECN parameter sample, an updated ECN inference model.
2. The method of claim 1 , further comprising performing, using the target ECN configuration parameter, congestion control in a second time period, wherein the second time period is later than the first time period in a time sequence.
3. The method of claim 1 , wherein the inference result further comprises a confidence of the original ECN configuration parameter, and wherein after obtaining the inference result and when the confidence is less than a confidence threshold, the method further comprises:
adjusting, based on a change of transmission performance of the network device, an ECN configuration parameter to obtain an adjusted ECN configuration parameter; and
using the adjusted ECN configuration parameter as the target ECN configuration parameter.
4. The method of claim 3 , wherein adjusting the ECN configuration parameter comprises:
when the transmission performance improves from a third time period to the first time period, increasing an ECN threshold in the ECN configuration parameter or lowering an ECN marking probability in the ECN configuration parameter, wherein the third time period is earlier than the first time period in a time sequence; and
when the transmission performance deteriorates from the third time period to the first time period, lowering the ECN threshold or increasing the ECN marking probability.
5. The method of claim 4 , wherein when a bandwidth utilization of the network device increases, a queue depth of the network device decreases, or an ECN packet ratio of the network device decreases from the third time period to the first time period, the method further comprises detecting that the transmission performance of the network device has improved from the third time period to the first time period.
6. The method of claim 1 , wherein the inference result further comprises a confidence of the original ECN configuration parameter, and wherein after obtaining the inference result and when the confidence is greater than or equal to a confidence threshold, the method further comprises using the original ECN configuration parameter as the target ECN configuration parameter.
7. The method of claim 1 , wherein the inference result further comprises a confidence of the original ECN configuration parameter, wherein after obtaining the inference result and when the confidence is less than a confidence threshold, the method further comprises sending target indication information to the analysis device, and wherein the target indication information comprises an identifier of the network device and indicates that the ECN inference model does not adapt to the network device.
8. The method of claim 7 , wherein the target indication information comprises the confidence.
9. The method of claim 1 , wherein after receiving the updated ECN inference model, the method further comprises updating, using the updated ECN inference model, the ECN inference model.
10. The method of claim 1 , wherein the first network status information comprises queue information, throughput information, or congestion information of the network device in the first time period.
11. A network device comprising:
one or more processors; and
one or more memories coupled to the one or more processors and configured to store programming instructions, wherein the one or more processors are configured to execute the programming instructions to cause the network device to:
input first network status information of the network device in a first time period into an explicit congestion notification (ECN) inference model;
obtain, based on the first network status information, an inference result from the ECN inference model, wherein the inference result comprises an original ECN configuration parameter;
send, to an analysis device that manages the network device, an ECN parameter sample comprising a target ECN configuration parameter, wherein the target ECN configuration parameter is based on the inference result; and
receive, from the analysis device and based on training with the ECN parameter sample, an updated ECN inference model.
12. The network device of claim 11 , wherein the programming instructions, when executed by the one or more processors, further cause the network device to perform, using the target ECN configuration parameter, congestion control in a second time period, and wherein the second time period is later than the first time period in a time sequence.
13. The network device of claim 11 , wherein the inference result further comprises a confidence of the original ECN configuration parameter, and wherein the programming instructions, when executed by the one or more processors, further cause the network device to:
adjust, based on a change of transmission performance of the network device, an ECN configuration parameter to obtain an adjusted ECN configuration parameter; and
use the adjusted ECN configuration parameter as the target ECN configuration parameter.
14. The network device of claim 13 , wherein the programming instructions, when executed by the one or more processors, further cause the network device to:
when the transmission performance improves from a third time period to the first time period, increase an ECN threshold in the ECN configuration parameter or lower an ECN marking probability in the ECN configuration parameter, wherein the third time period is earlier than the first time period in a time sequence; and
when the transmission performance deteriorates from the third time period to the first time period, lower the ECN threshold or increase the ECN marking probability.
15. The network device of claim 14 , wherein when a bandwidth utilization of the network device increases, a queue depth of the network device decreases, or an ECN packet ratio of the network device decreases from the third time period to the first time period, the programming instructions, when executed by the one or more processors, further cause the network device to detect that the transmission performance of the network device has improved from the third time period to the first time period.
16. The network device of claim 11 , wherein the inference result further comprises a confidence of the original ECN configuration parameter, and when the confidence is greater than or equal to a confidence threshold, the programming instructions, when executed by the one or more processors, further cause the network device to use the original ECN configuration parameter as the target ECN configuration parameter.
17. The network device of claim 11 , wherein the inference result further comprises a confidence of the original ECN configuration parameter, wherein when the confidence is less than a confidence threshold, the programming instructions, when executed by the one or more processors, further cause the network device to send target indication information to the analysis device, and wherein the target indication information comprises an identifier of the network device and indicates that the ECN inference model does not adapt to the network device.
18. The network device of claim 17 , wherein the target indication information comprises the confidence.
19. The network device of claim 11 , wherein the programming instructions, when executed by the one or more processors, further cause the network device to update, using the updated ECN inference model, the ECN inference model.
20. A computer program product comprising instructions stored on a non-transitory computer-readable medium that, when executed by one or more processors, cause a network device to:
input first network status information of the network device in a first time period into an explicit congestion notification (ECN) inference model;
obtain, based on the first network status information, an inference result from the ECN inference model, wherein the inference result comprises an original ECN configuration parameter;
send, to an analysis device that manages the network device, an ECN parameter sample comprising a target ECN configuration parameter, wherein the target ECN configuration parameter is based on the inference result; and
receive, from the analysis device and based on training with the ECN parameter sample, an updated ECN inference model.