Job scheduling based on historical job data

摘要

A method of job scheduling in a computing environment, comprising collecting historical job data related to job submissions received from a plurality of users, the historical job data comprising: resource request data indicative of resources requested by the user for executing a job, and resource usage data indicative of actual resources used to execute the job; calculating a believability score for each of the plurality of users based on a comparison between the job resource request data and the resource usage data related to job submissions received from the user, the believability score indicating a degree to which the user overestimates resources required to execute jobs; receiving a new job submission from a user, specifying a new job that is to be executed and resources requested by the user for executing the new job; and scheduling the new job for execution based on the believability score for the user.

主权项

1. A method of job scheduling in a computing environment, comprising:
collecting historical job data of job submissions received from a plurality of users, the historical job data comprising resource request data indicative of computing resources requested by each of the users for executing jobs, and resource usage data indicative of actual computing resources used by each of the users to execute the jobs; calculating a score for each of the plurality of users by comparing the resource request data and the resource usage data, the score indicating a degree to which a user overestimates computing resources required to execute the jobs; receiving a new job submission from a user of the plurality users, the new job submission specifying a new job to be executed and computing resources requested by the user for executing the new job; calculating a utilization factor for the new job being a ratio of the computing resources requested by the user for executing the new job to computing resources available during a backfill window; scheduling the new job for execution using the score of the user and the utilization factor of the new job; and executing the new job on a subset of the available computing resources during the backfill window, wherein the backfill window is a period of time in the future that the new job will execute and wherein the subset of the available computing resources comprises fewer computing resources than the computing resources requested by the user.