Edge machine learning system and method based on container cloud platform

Abstract

The invention discloses an edge machine learning system based on a container cloud platform. The edge machine learning system comprises an infrastructure subsystem, an edge routing subsystem, a storage subsystem, a machine learning computing equipment subsystem and a platform management subsystem. All the subsystems are containerized into edge container cloud platform service, resource scheduling,operation monitoring and the like are achieved in a service mode, and rapid deployment, operation and maintenance and release of the machine learning computing system based on the container cloud technology are achieved. According to the self-organizing cooperation method between the edge machine learning systems, distributed control and a multi-center network structure are achieved through interaction of the edge machine learning systems in a self-organizing communication mode, the identity correctness of the two communication parties is guaranteed through the access authentication technology, and it is guaranteed that important network propagation content is not stolen. The invention further provides a container inner volume collaborative migration method between the edge machine learning systems, data between the container cloud platforms can be migrated mutually through container inner volume collaborative migration, and service integrity and inertness are guaranteed.

Description

technical field [0001] The present invention belongs to the related field of container cloud technology, and in particular relates to an artificial intelligence edge computing (referred to as edge artificial intelligence) system and method based on container cloud technology. Background technique [0002] With the rise of artificial intelligence machine learning algorithms, there is a greater need for data storage and migration. The research and development and improvement of artificial intelligence algorithms need to repeatedly call the download and upload of data calculated in the order of TB, which brings high cost and bandwidth requirements to the traditional transmission network. At the same time, the high cost of artificial intelligence computing equipment also requires effective scheduling and allocation of artificial intelligence computing equipment. [0003] With the rise of edge artificial intelligence, it has transformed from centralized to distributed, and based...

AI Technical Summary

Problems solved by technology

[0004] The current deployment mode of edge artificial intelligence is based on the traditional server deployment form, relying on the underlying infrastructure, and does not have the advantages of efficient and simple deployment and distribution of services, high process availability, and automatic monitoring of service running status, and cannot automatically pull up downtime

Images