Cloud application container resource scheduling method based on feed-forward control

Abstract

The invention relates to a cloud application container resource scheduling method based on feed-forward control. The method comprises the following steps: a plurality of features of micro-service architecture are firstly extracted; then, incidence relation of work load, response time and resource utilization rate is established through an earnings equation of a Jackson open-loop network; parameters of the response time, the load and the like are predicted by adopting a fuzzy self-adaptive kalman filtering algorithm, so that timeliness of resource supply is improved, and quality of applicationservice is guaranteed; support policies of task type, task priority, server resource utilization rate and the like are added on the premise that the response time not exceeding the quality of the application service is taken as a primary scheduling policy; three scheduling methods of extension, shrinking and migrating are provided, so that container-level multi-task hybrid deployment is realized.

Description

technical field [0001] The invention relates to a cloud application container resource scheduling method based on feedforward control, which belongs to the field of software technology. Background technique [0002] In the field of software design and development, the microservice architecture can well meet the needs of fast-changing Internet application demands and independent update of modules due to its strong module autonomy. However, in the field of software operation and maintenance, because the number of modules is much higher than that of the traditional architecture, it undoubtedly increases the cost and difficulty of operation and maintenance. Therefore, how to provide adaptive resources has become the main problem that the microservice architecture needs to solve urgently. one of the problems. The essence of a container is a process group that holds independent resources and is isolated through a namespace. It has the characteristics of less resource consumption ...

AI Technical Summary

Problems solved by technology

[0004] The existing work generally has the following three deficiencies: 1) The characteristics of a large number of microservice architecture modules are not considered, and each service module is modeled separately, resulting in a large cost of the provisioning method; 2) The incremental resources of the container are not considered Due to the characteristics of use, the resource supply is based on a fixed threshold as a standard, resulting in poor supply timeliness; 3) The irregular characteristics of sudden loads are not considered, and modeling is based on a large amount of historical data, resulting in low supply accuracy.

Images