Method and device for resource scheduling

Abstract

The purpose of this application is to provide a method and device for resource scheduling, which includes: obtaining the resource usage status of the private resource pool; if the resource of the private resource pool exceeds the limit, requesting the resource allocation device to block the user's Private resource pool resource application, transfer all or part of the work of the task that currently causes the private resource pool resources to exceed the limit to the public resource pool, after the transfer is completed, request the resource allocation device to stop blocking the private resource pool resource application .

Description

technical field [0001] The present application relates to the field of computers, and in particular to a technology for resource scheduling. Background technique [0002] Resource scheduling is an essential basic function of cloud computing, and the effect of resource scheduling directly determines the stability and cost of cloud computing. There are usually many resources that need to be isolated, such as CPU (central processing unit), memory, IO (input / output), etc. So far, there are many resource scheduling methods developed. In order to maximize resource utilization, generally in Multiple tasks run on one machine, but if we limit resources, the size of the limited resources directly determines the life and death of the task, such as the memory isolation limit. If the task exceeds the memory limit, it will trigger OOM (Out of Memory, memory overflow), the task will be killed gracefully. For a user of a cloud product, the killing of a task will obviously affect the opera...

AI Technical Summary

Problems solved by technology

There are usually many resources that need to be isolated, such as CPU (central processing unit), memory, IO (input / output), etc. So far, there are many resource scheduling methods developed. In order to maximize resource utilization, generally in Multiple tasks run on one machine, but if we limit resources, the size of the limited resources directly determines the life and death of the task, such as the memory isolation limit. If the task exceeds the memory limit, it will trigger OOM (Out of Memory, memory overflow), the task will be killed gracefully

For a user of a cloud product, the killing of a task will obviously affect the operation of the business, and the current common practice is to restart the task, but it still makes the user feel uncomfortable, because a task is often multiple processes or threads , the original resource overrun caused by one process or thread also caused other processes or threads to suffer

Images