Data distribution method, medium and electronic device

Abstract

The present invention provides a data distribution method, medium, and electronic equipment. The method includes: a node with data to be distributed runs a push program, and the push program is used to repeatedly perform push processing before the end of the entire distribution process. The push The processing includes: pushing the data to a node to be pushed according to the transmission rate between the nodes, wherein the transmission rate from the node with the data to be distributed to the node to be pushed is lower than that between the node with the data to be distributed and all the nodes to be pushed The transmission rate of the push node is the highest and not 0; the node to be pushed runs the push program as a node with the data to be distributed after receiving the data. The invention reduces the time delay of data distribution in the network and effectively improves the utilization rate of network bandwidth.

Description

technical field [0001] The present invention relates to the technical field of data sharing, in particular to a data distribution method, medium and electronic equipment. Background technique [0002] As a kernel-level virtualization technology, Docker provides an isolated environment for each container to run applications on the same machine. Developers can easily customize and package the runtime environment of their applications in a set of files, which are called Docker images (hereafter referred to as images). Once a Docker image is created, users can run it in any The image runs on the host (or node), which simplifies application development, delivery, and deployment. However, in project practice, it is sometimes necessary to distribute a container to a large number of nodes at the same time. At this time, if the local Docker mirror warehouse is used, the deployment time will be too long, and as the bandwidth is exhausted, the image service provided by the Docker mirr...

AI Technical Summary

Problems solved by technology

However, the above methods ignore the bandwidth difference between nodes, and will occupy a large amount of bandwidth during distribution, which will cause huge network pressure on other high-real-time applications. On the one hand, the bandwidth of the entire network will be completely occupied at the beginning of the distribution task. , and the occupied bandwidth can only be released after the distribution task ends; on the other hand, in the case of transmitting the same file, nodes with low bandwidth will take longer to transmit than nodes with high bandwidth, thus prolonging the overall distribution schedule

Images