Distributed system scheduling method and device and electronic equipment

Abstract

The invention discloses distributed system scheduling method and device and electronic equipment. The distributed system scheduling method includes that a management center of a distributed system acquires quantity and sequence number M of resource providing units of the distributed system; the management center of the distributed system acquires quantity and number of resource consumption units; the quantity of the resource consumption units is divided by the quantity of resource providing units to obtain the resource consumption unit average bearing quantity N born by the resource providing units; N or (N-1) or (N+1) resource consumption units are extracted from the resource consumption units according to the sequence of the number of the resource consumption units and are taken as the resource consumption units distributed to the resource providing units on the basis of the resource providing unit sequence number M. According to the technical scheme, problems of low task handling efficiency and deadlock possibility caused during task distribution carried out by means of distributed locks of the distributed system, and low speed of convergence in task distribution and long distribution time caused by circular preemption are solved.

Description

technical field [0001] This application relates to the field of computer technology, and specifically relates to two scheduling methods for distributed systems; corresponding to the above methods, this application also relates to two scheduling devices for distributed systems and two electronic devices. Background technique [0002] Today, with the continuous development of Internet technology, more and more data processing relies on high computing power. Use distributed systems to deal with the above problems. [0003] In a distributed system, in order to allow various systems to work together, the concept of distributed locks is proposed. Distributed locks are a way to control access to shared resources between distributed systems. Threads in machines that grab distributed locks , you can continue to execute the task, and other machines wait for the lock to be released. [0004] It can be seen that because the distributed lock has a process of grabbing locks and releasin...

AI Technical Summary

Problems solved by technology

[0004] It can be seen that because the distributed lock has a process of grabbing locks and releasing locks, and only one machine can grab the lock at the same time, the machine that grabs the lock can run, so the efficiency of the machine in processing tasks is low, and the machine In the process of using distributed locks, if the machine that grabs the lock suddenly crashes and the lock has not been released, it will cause deadlock

In addition, when preempting, if the machine fails to preempt once, it needs to preempt in a loop, so the convergence of task allocation is slow and the allocation time is long

Images