A fault-tolerant task scheduling method in virtualized cloud

Abstract

The invention discloses a fault-tolerant task scheduling method in a virtualized cloud, which is characterized in that the information of the arrived dependent task group and the physical host information of the virtualized cloud are obtained; the PB model is used to establish a master task for each task in the dependent task group version and minor version; specify an earliest start time and a latest completion time for each version of each task in the dependent task group; divide multiple virtual machines on each activated physical host, and obtain each The information of each virtual machine on the activated physical host; each version of each task in the dependent task group will be loaded to each activated virtual machine in the specified time period according to the distribution of tasks on each activated physical host In each virtual machine on the activated physical host; run each version of each task in the loaded dependent task group according to the specified time schedule; complete all tasks of the dependent task group and return the task result.

Description

technical field [0001] The invention relates to the field of cloud computing, in particular to a fault-tolerant task scheduling method in a virtualized cloud. Background technique [0002] Due to the unpredictability of computer system errors, it is very important to add support for fault tolerance when designing scheduling algorithms. Fault-tolerant scheduling algorithms can be roughly divided into two categories, namely static fault-tolerant scheduling and dynamic fault-tolerant scheduling: static fault-tolerant scheduling makes scheduling decisions before task submission, and is usually used to schedule periodic tasks; dynamic fault-tolerant scheduling is usually used to schedule non-periodic tasks , whose task arrival time is uncertain. [0003] At present, there are two main fault-tolerant scheduling methods in the distributed computing environment, namely, resubmission and replication. Resubmission means that when the computing node to which a task is assigned fails,...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the lack of a fault-tolerant task scheduling method in the cloud computing environment in the prior art, the purpose of the present invention is to propose a fault-tolerant task scheduling method in a virtualized cloud, which can use the PB model to schedule fault-tolerant tasks in a cloud computing environment, improving Resource utilization and schedulability of fault-tolerant tasks

Images