Job management based on generated execution time prediction model and prediction accuracy

Patent metadata

Field	Value
Publication number	US-11734063-B2
Application number	US-202117198730-A
Country	US
Kind code	B2
Filing date	Mar 11, 2021
Priority date	Jun 3, 2020
Publication date	Aug 22, 2023
Grant date	Aug 22, 2023

Abstract

Execution end timing of a jobnet is predicted with stable accuracy. An information processing apparatus executes jobnet execution time prediction model generation processing for generating an execution time prediction model for predicting execution time of a jobnet to be executed on the basis of information associated with execution time of each of previously executed jobnets, a prediction model accuracy determination processing for calculating prediction accuracy for the execution time of each of jobnets by the generated execution time prediction model, and delay determination processing for determining whether to predict execution end timing of a designated jobnet among a jobnet group that is a set of a plurality of jobnets currently being executed or to be subsequently executed on the basis of the execution time prediction model, on the basis of the calculated prediction accuracy for the execution time of each of the jobnets.

First claim

What is claimed is:

1. A job management method performed by a job management system comprising a processor and a memory, the job management method comprising the steps of:
jobnet execution time prediction model generation processing for:
acquiring information about a resource usage condition of the job management system at a time of previously executing each jobnet of the plurality of jobnets; and
generating an execution time prediction model for predicting execution time for each jobnet based on the acquired information associated with execution time of each of the previously executed jobnets;
prediction model accuracy determination processing for calculating prediction accuracy for the execution time of each of the plurality of jobnets by the generated execution time prediction model;

jobnet execution time collection processing for calculating a statistical value of the execution time of each of the plurality of jobnets on a basis of the information associated with the execution time of each of the previously executed jobnets;
delay determination processing for determining whether to predict execution end timing of a designated jobnet among a jobnet group that is a set of a plurality of jobnets currently being executed or to be subsequently executed on a basis of the execution time prediction model, on a basis of the calculated prediction accuracy for the execution time of each of the plurality of jobnets, and for displaying information associated with the execution end timing of the designated jobnet predicted on the basis of the execution time prediction model in a case of determining to predict the execution end timing of the designated jobnet on the basis of the execution time prediction model,
wherein the delay determination processing further includes:
determining whether to predict the execution end timing of the designated jobnet on the basis of the execution time prediction model by determining whether the prediction accuracy for the execution time of the designated jobnet by the execution time prediction model is equal to or higher than a predetermined value; and
displaying the information associated with the execution end timing of the designated jobnet predicted on the basis of:
(a) the execution time prediction model in a case of determining that the prediction accuracy for the execution time is equal to or higher than the predetermined value; or
(b) the calculated statistical value in a case of determining that the prediction accuracy for the execution time is not equal to or higher than the predetermined value; and

executing the designated jobnet on the basis of the execution time prediction model.

2. The job management method according to claim 1 ,
wherein the delay determination processing includes waiting for a predetermined period of time before displaying the information associated with the execution end timing of the designated jobnet calculated based on the calculated statistical value in the case of determining that the prediction accuracy for the execution time is not equal to or higher than the predetermined value.

3. The job management method according to claim 1 ,
wherein the jobnet execution time collection processing includes calculating, as the statistical value of the execution time of each of the jobnets, an average value and a maximum value of the execution time of each of the jobnets on the basis of the information associated with the execution time of each of the previously executed jobnets,
wherein the delay determination processing includes further determining whether the execution time of the designated jobnet is equal to or longer than predetermined time in the case of determining that the prediction accuracy for the execution time is not equal to or higher than the predetermined value, displaying the information associated with the execution end timing of the designated jobnet calculated based on the calculated maximum value in a case of determining that the execution time of the designated jobnet is equal to or longer than the predetermined time, and displaying the information associated with the execution end timing of the designated jobnet calculated based on the calculated average value in a case of determining that the execution time of the designated jobnet is not equal to or longer than the predetermined time.

4. The job management method according to claim 1 , wherein the jobnet execution time prediction model generation processing includes estimating an order of executing the jobnets on a basis of the information associated with execution time of each of the previously executed jobnets, and
wherein the delay determination processing includes identifying a jobnet earlier than the designated jobnet in the order of executing the jobnets on a basis of the estimated order of executing the jobnets.

5. The job management method according to claim 1 ,
wherein the jobnet execution time prediction model generation processing includes calculating, as one piece of the information associated with the execution time of each of the previously executed jobnets, a wait time that is a difference between an execution start time of each of the previously executed jobnets and an execution end time of other jobnet execution of which is ended before execution of each of the jobnets, and generating the execution time prediction model on a basis of the calculated wait time.

6. A job management apparatus comprising:
a processor; and
a memory,
the job management apparatus configured to execute:
jobnet execution time prediction model generation processing for
acquiring information about a resource usage condition of the job management apparatus at a time of previously executing each jobnet of the plurality of jobnets; and
generating an execution time prediction model for predicting execution time for each jobnet based on the acquired information associated with execution time of each of the previously executed jobnets,

prediction model accuracy determination processing for calculating prediction accuracy for the execution time of each of the plurality of jobnets by the generated execution time prediction model,
jobnet execution time collection processing for calculating a statistical value of the execution time of each of the plurality of jobnets on a basis of the information associated with the execution time of each of the previously executed jobnets;
delay determination processing for determining whether to predict execution end timing of a designated jobnet among a jobnet group that is a set of a plurality of jobnets currently being executed or to be subsequently executed on a basis of the execution time prediction model, on a basis of the calculated prediction accuracy for the execution time of each of the plurality of jobnets, and for displaying information associated with the execution end timing of the designated jobnet predicted based on the execution time prediction model in a case of determining to predict the execution end timing of the designated jobnet on the basis of the execution time prediction model,
wherein the delay determination processing includes:
determining whether to predict the execution end timing of the designated jobnet on the basis of the execution time prediction model by determining whether the prediction accuracy for the execution time of the designated jobnet by the execution time prediction model is equal to or higher than a predetermined value; and

displaying the information associated with the execution end timing of the designated jobnet predicted on the basis of:
(a) the execution time prediction model in a case of determining that the prediction accuracy for the execution time is equal to or higher than the predetermined value; or
(b) the calculated statistical value in a case of determining that the

Assignees

• Hitachi Ltd

Inventors

• Kitawaki Jun
• Tameshige Takashi
• Tamai Yasuyuki
• Murayama Kouichi
• Masuda Mineyoshi
• Himura Yosuke

Classifications

• G06F9/4881 Primary

  Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues · CPC title

• G06F9/5038

  considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration (scheduling strategies G06F9/4881 and subgroups) · CPC title

• G06F9/4887 Primary

  involving deadlines, e.g. rate based, periodic · CPC title