Cluster task scheduling system

Abstract

The present application discloses a clustered task dispatching system, which is characterized by the processing of task dispatching. The clustered task dispatching system of the present application includes a command dispatching platform, a JVM container and a node server. The command dispatching platform is composed of multiple task servers connected to each other through the network. Zookeeper services are deployed on these task servers, and the zookeeper service returns execution commands according to task requirements. Each JVM container is used to run business services, and different running instructions correspond to JVM containers and business services. The node server receives the running command from the command dispatching platform or returns and sends the command to the command dispatching platform. The node server judges whether the load status of the JVM container that obtained the running command is idle or busy. If the load status is idle, the JVM container processes the task requirements according to the running command. If the load status is busy, the node server temporarily stores the running command in the waiting queue. .

Description

technical field [0001] This application relates to a network service task dispatching system, in particular to a clustered task dispatching system. Background technique [0002] With the high-speed improvement of network speed, it also drives the diversity of network services. Behind the network service is realized by the collocation of hardware and software. The hardware aspect refers to the architecture, resources or computing power of the server. In terms of software, it is realized by calculation mechanisms such as task allocation and resource scheduling. [0003] At present, the more popular task scheduling platforms include Quartz-based, Redis-based storage of task queue data, use of crontab commands, Mapduce-based computing framework, IBM MQ message queue, etc. These scheduling platforms all have some flaws: Quartz’s stand-alone version of single-process applications cannot ensure that tasks are triggered correctly in case of downtime, and there is a risk of memory...

AI Technical Summary

Problems solved by technology

[0005] Compared with the prior art, this application is based on the Zookeeper distributed application coordination service, and can obtain the following technical effects for the ever-increasing number of tasks, more and more detailed task content, and more and more complex task management :

Images