Priority queue based cloud task multidimensional constraint backfill scheduling method

Abstract

The invention discloses a priority queue based cloud task multidimensional constraint backfill scheduling method. The method overcomes the disadvantages of poor backfill performance caused by the reason that consideration of task indexes in a traditional backfill scheduling algorithm and an improved backfill scheduling algorithm in the cloud computing is too single and an enlarged resource duty ratio of a virtual machine. The method comprises the steps of balancing by giving comprehensive consideration to the applying number of the processor cores and a preset execution duration of an independent task in a batch task; building a minimum binary reactor based priority queue by utilizing a value of a backfill weight coefficient to preferentially backfill a task with less applied resources and relatively long preset execution time; and adding constraint conditions such as memory occupation and bandwidth occupation. The priority queue based cloud task multidimensional constraint backfill scheduling method improves the utilization rate of computing resources of a virtual machine in unit time, shortens the final finishing time of the batch task and reduces the average waiting delay of tasks.

Description

technical field [0001] The present invention relates to the technical field of cloud computing, specifically applied to resource management technology in cloud computing platforms, and in particular to a priority queue-based cloud task multi-dimensional constraint backfill scheduling method. Background technique [0002] Today, we are in the era of information explosion, and the development of computer technology is changing with each passing day. The current trend is a sharp increase in the number of users, and traditional data centers are no longer able to meet the existing needs. Then brought the wide application of cloud computing, a new technology, and its birth is also a product of the information technology revolution. Cloud computing is usually oriented to heterogeneous components, and in the face of massive task requests, how does the cloud computing data center schedule and process tasks in the shortest time within limited computing resources, shorten the final co...

AI Technical Summary

Problems solved by technology

But in this way, it is easy to prioritize the scheduling of tasks with small execution time, while the response time of tasks with long execution time becomes larger

Unable to balance the scheduling relationship between tasks of different lengths

[0007] In addition, some traditional backfill algorithms and existing improved backfill scheduling methods only consider the single index of the number of processor cores waiting for task applications when selecting backfill tasks, but do not combine the requested resources with the estimated execution time of independent tasks. In combination, this single consideration method can only be effective when the execution time of independent tasks is similar. Once the execution time of independent tasks is greatly different, the degree of idle hunger of the virtual machine will increase, resulting in longer task completion time. , the average waiting delay of tasks increases

Moreover, in the existing task scheduling methods specific to the cloud computing environment, there is no efficient method based on backfill scheduling that comprehensively considers the constraints of multi-dimensional performance indicators.

Images