Dynamic orchestration and task scheduling method and system for edge cloud hybrid job

Abstract

This invention discloses a dynamic orchestration and task scheduling method and system for edge-cloud hybrid operations, comprising: In each time window, the cloud predicts the arrival rate of online service requests and obtains a job orchestration set under the condition of not exceeding the total communication capacity and total computing power of each edge cluster; in each time slot, the cloud obtains a request scheduling set by solving a linear programming problem with the objective of maximizing the number of online service requests; when a job request arrives at an edge cluster, the edge cluster judges the request. If it is an online service, the edge cluster performs deployment orchestration based on the job orchestration set and the request scheduling set, or sends it to other edge clusters; if it is an offline job, the edge cluster forwards the request to the cloud, and the cloud generates a job scheduling set based on a weighted scoring method of comprehensive indicators, and the edge cluster processes the offline batch processing job. This invention can improve the total throughput and resource utilization of the system while ensuring the service quality of online services.

Images