Task scheduling method based on distributed memory clusters

Abstract

The invention relates to the technical field of task scheduling, in particular to a task scheduling method based on distributed memory clusters. The method includes: adding a created task into a task queue to wait for loading; adding one task in the task queue to a distributed memory each time; knowing utilization conditions and utilization rates of task machines by the aid of a monitoring module, and knowing execution conditions of each task by the aid of the monitoring module; allocating the task loaded each time into the task machine lower in utilization rate. By adoption of the task scheduling method based on the distributed memory clusters, dynamically transverse expansion and shrinkage can be realized without affecting normality in system operation, and timeliness in control of task execution conditions and modification of task related parameters is realized; tasks can be allocated reasonably to prevent each machine from overloaded task execution; a complete system monitoring interface is provided to enable people to timely know operation conditions of a whole system; the system is high in scheduling reliability and stability.

Description

technical field [0001] The invention relates to the technical field of task scheduling, in particular to a task scheduling method based on distributed memory clusters. Background technique [0002] In recent years, with the rapid development of the Internet field, the scale of task machines (servers) on which the Internet depends has also become larger and larger. And with the continuous expansion of the application field, multiple applications of the user are often executed on different task machines. Therefore, how to manage, maintain and schedule these cross-machine application service relationships has become an urgent problem to be solved. [0003] For a large number of complex tasks, the configuration of the vertical lifting computer can no longer meet the operational requirements of the task. Horizontal scaling is the best way to solve the problem. The following problems exist in the existing distributed scheduling system: [0004] 1. Tasks cannot be allocated rea...

AI Technical Summary

Problems solved by technology

[0004] 1. The tasks cannot be allocated reasonably according to the overall performance of the machine, and they are all competitive or equal allocation tasks

[0005] 2. Cannot achieve real-time control tasks

Many framework designs are based on the fact that after the task is assigned to the machine, operations such as deleting, modifying, and checking cannot be performed on it during its operation.

[0006] 3. Cannot automatically monitor the entire system, monitor the running status of tasks and the task carrying capacity of the entire system

Images