Method of predicating time duration

Abstract

The invention discloses a method of predicating time duration. The method comprises the following steps: acquiring waiting time duration of to-be-executed tasks in a distributed system and execution time duration of executing the to-be-executed tasks, wherein the waiting time duration of one task is a time duration interval between the current time and the execution starting time of one task, and the execution time duration of one task is a time duration interval between the execution starting time and the execution finishing time of one task; acquiring residual time duration of the to-be-executed tasks according to the waiting time duration and execution time duration of the to-be-executed tasks, wherein the residual time duration of one task is a time duration interval between the current time and the execution finishing time of one task. By applying the method disclosed by the invention, the residual time duration of the to-be-executed tasks can be obtained, i.e., the time duration interval between the current time and the execution finishing time of each to-be-executed task is obtained, and thus subsequent resource scheduling and other task arrangement can be facilitated.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a method for estimating the duration of a processing task. Background technique [0002] With the advent of big data, distributed computing has become the main way to solve big data. In a distributed system, a large task is generally decomposed into multiple small tasks for concurrent processing, and then the processed results are combined and output. For example, in the distributed cloud transcoding platform, to convert a source file into video output in other formats, the source file is first cut into several small pieces, and then dispatched to the transcoder group for concurrent transcoding. Merge output videos; however, the number of transcoder groups is a fixed value, which is the maximum number of transcoding tasks that can be transcoded at the same time. When the number of submitted tasks exceeds this value, the tasks at the front of the queue will be processed first. A f...

AI Technical Summary

Problems solved by technology

Due to the influence of factors such as the number of concurrent tasks, processor resource utilization, processor processing speed, scheduling algorithm, and the complexity of the task itself, there is no one that can estimate the remaining time of task processing in the current distributed environment. Methods

Therefore, the distributed system cannot provide users with the remaining time for the distributed system to complete a task in a timely manner, which makes it difficult for subsequent resource scheduling or other task scheduling plans to proceed smoothly.

Images