Container cloud-oriented task collaborative scheduling system and method thereof

Abstract

The invention provides a container cloud-oriented task collaborative scheduling system and method thereof. An environment feature extraction module is used for extracting feature indexes required by task collaborative scheduling from a current resource environment, and supporting a scheduling scheme construction module to carry out the judgment of the node scheduling capability; the scheduling scheme construction module is used for constructing an optimal scheduling scheme in the current environment for the current to-be-scheduled task; the scheduling scheme execution module is used for executing a scheduling scheme, namely, scheduling a task to be scheduled to an optimal scheduling node specified by the scheduling scheme; and the scheduling evaluation and feedback adjustment module is used for evaluating the actual scheduling effect of the optimal scheduling scheme, that is, the improvement condition of the environment resource state after task scheduling is judged according to an evaluation rule, and the judgment result is fed back to the scheduling scheme construction module for improving the scoring accuracy of the scoring algorithm. According to the invention, the resource convergence effect and the scheduling performance are improved in the aspect of multi-target task scheduling with different dependency feature relationships.

Description

technical field [0001] The invention relates to a container cloud task scheduling technology, in particular to a container cloud-oriented task collaborative scheduling system and a method thereof. Background technique [0002] As a mainstream branch of cloud computing development, container cloud further refines cloud computing, enabling it to provide high-quality computing resources in a lighter, more economical, and faster manner. With the continuous development of container cloud technology, more and more applications have begun to migrate to the cloud after containerized transformation, or directly face the container cloud, using cloud-native architecture for development, integration and deployment, relying on lightweight containers, low resource consumption, and start-up Fast, and the cloud's powerful centralized storage and processing capabilities and other advantages, aiming to provide users with more efficient and high-quality tasks. However, due to the architectura...

AI Technical Summary

Problems solved by technology

At present, the bottom layer of most container cloud platforms uses a centralized scheduling architecture. Although this architecture design achieves the consistency of underlying resources, it is difficult to optimize the combination of global resources in a single-scenario scheduling model, and it is also difficult to consider more from the business level. Useful Information for Participating in Scheduling

However, early task scheduling algorithms such as greedy algorithm, heuristic algorithm, and hierarchical analysis algorithm generally complete scheduling actions from the perspective of inherent resource usage, lacking interactive feedback with the actual operating environment, especially for multi-copy and multi-dependent task scheduling, which is prone to fall into local optimum Excellent, unable to achieve optimal convergence

To sum up, there is currently a lack of collaborative scheduling methods and systems for container cloud tasks in related fields, so as to further improve the resource utilization and task quality of container clouds, and realize optimal scheduling of global resources on container cloud platforms.

Images