Distributed system task execution method and device

Abstract

The invention provides a distributed system task execution method and device. A control task stream and a result data stream can be separated to solve the system bottleneck problem of a task scheduler, the system resource consumption is reduced, and the system reliability is improved. The distributed system task execution method comprises the following steps: a task initiator creates a task and sends the task to the task scheduler; the task scheduler receives the task and distributes the task to a task processor; the task processor receives and executes the task, and saves an execution resultof the task; and the task initiator acquires the execution result of the task from the saved execution result.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method and device for executing distributed system tasks. Background technique [0002] A distributed task processing system is a computer system that combines multiple autonomous and interconnected computers to complete the same task under the coordination of a high-level operating system in order to improve the parallel processing speed of tasks. [0003] The existing distributed task processing system takes Gearman as an example, such as figure 1 Shown is the block diagram of the implementation of the existing Gearman task distribution system. The processing of a Gearman task involves three roles: Client, JobServer and Worker. in: [0004] Client, the initiator of the task, creates the task that needs to be executed and then sends it to the task scheduler JobServer; [0005] JobServer, the task scheduler, is responsible for coordinating and forwarding the task ...

AI Technical Summary

Problems solved by technology

[0009] 1. JobServer is the node with the heaviest load in the entire system, and the performance of a single machine will seriously affect the performance of the entire system;

[0010] 2. The communication between Client, JobServer, and Worker will consume a lot of system resources. When the Client needs to obtain the result data multiple times, it will bring a huge system burden;

[0011] 3. The reliability of the system is low. If a Worker fails, it will affect the acquisition of historical data running on the Worker

Images