Data transmission method and network equipment

Abstract

The invention discloses a data transmission method and network equipment; the requirements of a data packet on latency and throughput rate are met under the condition of multiple queues. Firstly, a first data packet sent by a data sending end is received; according to the service type of the first data packet, the first data packet is determined to enter a first queue of the N queues for scheduling; the N queues are used for transmitting a data packet carrying an explicit congestion signal; the N queues correspond to different priorities respectively; the priorities are determined according tothe Latency and the Throughput; when the first data packet enters the first queue, and when the used length of the first queue exceeds a marking threshold value corresponding to the first queue, a congestion mark is added on the explicit congestion signal carried in the first data packet to obtain a second data packet; the first queue is scheduled to send the second data packet to a data receiving end according to the priority corresponding to the first queue; and the data receiving end sends a response message carrying the congestion mark to the data sending end after receiving the second data packet.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a data transmission method and network equipment. Background technique [0002] With the development of technologies such as cloud computing, big data, and the Internet of Things, there are many emerging services in the current network, and behind the good experience of these services are the requirements for throughput and latency. Therefore, the network needs It can support the different requirements of different services on throughput and delay at the same time. [0003] However, the actual situation is that, for networks currently using Transmission Control Protocol (TCP) as the main transmission protocol, when the bandwidth and network topology are fixed, queues are needed to support concurrency and reduce packet loss, while TCP itself This feature makes it difficult to ensure high throughput and low latency at the same time in the presence of queues. This is due ...

AI Technical Summary

Problems solved by technology

This creates a dilemma that improving throughput requires deep buffers, while reducing latency requires shallow buffers.

[0006] In summary, due to the existence of buffers in the prior art, high throughput and low latency cannot be satisfied at the same time.

In addition, the prior art usually optimizes management for a single queue, such as using the Reno scheme for management. This scheme will blindly reduce the sending window by a large proportion based on packet loss, resulting in severe jitter in the throughput rate.

In network devices such as routers and commercial switches, there are generally 4-8 queues. In the case of multiple queues, multiple queues still need to share a port. Insufficient forwarding performance, queue congestion, and increased delay

Images