Job Scheduling Control Method for Distributed Cloud Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed information processing systems, the execution of jobs across multiple information processing systems is hindered by increased data transfer times between cloud services, leading to prolonged job execution times due to inefficient data input/output operations, especially when data updates occur between evaluation and execution times.
Innovation Solution
A control method that identifies an optimal time period for evaluating data amounts and determines the most suitable system for job execution based on the evaluated data, ensuring that the system with the largest data amount is chosen for processing, thereby minimizing data transfer delays and optimizing job execution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If data amount evaluation is performed at a fixed timing before job execution, then the job execution system can be determined in advance, but the evaluated data amount may become outdated due to data updates between evaluation and execution times
Solution Approach 1:
The patent applies dynamics by making the evaluation timing adaptive rather than fixed. The control device dynamically determines evaluation timing based on the actual execution timing of jobs, ensuring that evaluation occurs when data updates are least likely to have occurred. This resolves the contradiction by allowing the system to adapt evaluation timing to actual system conditions, maintaining both timely execution and accurate data amount evaluation.
Solution Approach 2:
The patent implements feedback by using execution history information to identify optimal evaluation timing. The control device continuously monitors job execution patterns and uses this feedback to determine when data updates are minimal, thereby selecting the best timing for data amount evaluation. This feedback mechanism ensures that evaluation accuracy is maintained while supporting efficient job execution.
2Productivity
If data transfer between cloud services is increased to enable job execution across multiple systems, then processing capability is improved, but data transfer time increases leading to prolonged job execution
Solution Approach 1:
The patent extracts only the necessary data for job execution from the distributed cloud services, rather than transferring all available data. By identifying and transferring only the specific data items required for the current job, the system maintains processing capability across multiple systems while minimizing data transfer time and improving overall efficiency.
3Reliability
If data updates are allowed to occur continuously in the data storage area, then data freshness is improved, but data amount evaluation becomes inaccurate when performed at fixed intervals
Solution Approach 1:
The patent makes the evaluation timing dynamic by adjusting it based on job execution patterns and data update frequencies. Instead of using fixed evaluation intervals, the system identifies optimal timing points where data updates are minimal, thereby maintaining both data freshness for ongoing operations and evaluation accuracy for job scheduling decisions.
Solution Approach 2:
The patent performs data amount evaluation at strategically chosen timing points before job execution, when data updates are least likely to occur. This preliminary evaluation at optimal timing ensures accurate data amount assessment while still allowing continuous data updates to occur for maintaining data freshness during normal operations.
Data Source
AI summary
A control method executed by a control device, the control method including identifying a specified time period based on execution history information on previous jobs related to the plurality of systems, the specified time period being a period prior to execution start timing of a first job, update processing for a data storage area from which the first job refers to data is not executed during the specified time period, performing control so that evaluation timing of an amount of the data that the first job refers to from the data storage area is included in the identified time period, determining a specified system among from the plurality of systems based on the amount of the data evaluated at the evaluation timing, and causing the specified system to execute the first job.