Dynamic scheduling method, apparatus, and system

A scheduling method, comprising:

• determining, by an aggregation switch, first virtual output queue (VOQ) length information in each timeslot, wherein the aggregation switch is connected to n/2 core switches, wherein n is equal to or grater than 2, where n is an integer, wherein the first VOQ length information comprises a quantity of data packets buffered in M-1 VOQs in the aggregation switch, wherein each of the M-1 VOQs is used to buffer a data packet that needs to be sent from a first cluster pod to one of M-1 second pods, wherein the first cluster pod is a pod to which the aggregation switch belongs, wherein the M-1 second pods are other pods than the first pod in M pods managed by a scheduling platform, wherein M is equal to or greater than 2, and wherein M is an integer;
• sending, by the aggregation switch, the first VOQ length information to the scheduling platform;
• receiving, by the aggregation switch, matching result information sent by the scheduling platform, wherein the matching result information is used to instruct to send a first data packet, wherein the first data packet is a data packet that needs to be sent from the first pod to a third pod, and wherein the third pod is one of the other pods;
• and
• sending, by the aggregation switch, the first data packet based on the matching result information.

A scheduling platform, comprising:

• a communications interface, configured to receive first virtual output queue (VOQ) length information of each aggregation switch in each timeslot, wherein the first VOQ length information comprises a quantity of data packets buffered in M-1 VOQs in the aggregation switch, wherein each of the M-1 VOQs is used to buffer a data packet that needs to be sent from a first cluster pod to one of M-1 second pods, wherein the first cluster pod is a pod to which the aggregation switch belongs, wherein the M-1 second pods are other pods than the first pod in M pods managed by the scheduling platform, wherein M is equal to or greater than 2, and wherein M is an integer;
• a processor, configured to aggregate the first VOQ length information received by a receiving unit to obtain global VOQ length information, wherein the global VOQ length information comprises a total quantity of data packets that need to be sent from each of the M pods to other pods than the pod in the M pods;
• wherein the processor is further configured to determine a transmission matching scheme based on the global VOQ length information, wherein the transmission matching scheme includes is determined by converting the global VOQ length information into a weighted graph, and obtaining the transmission matching scheme from the weighted graph using a weighted matching algorithm, wherein the transmission matching scheme is used to instruct each pod to send a data packet to one of the M pods except the pod and receive a data packet sent by one of the M pods except the pod;
• and
• wherein the communication interface is configured to send corresponding matching result information to each aggregation switch according to the transmission matching scheme determined by the processor, so that each aggregation switch sends a data packet according to the transmission matching scheme.