Job Scheduling for Parallel Nodes with Version Mismatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In parallel processing systems, the initiation of update jobs on different nodes at different times can lead to long delays in the execution of subsequent user jobs, affecting utilization efficiency and waiting times.
Innovation Solution
A job scheduling method that identifies update jobs and user jobs with specified node usage and execution times, plans a schedule to minimize total waiting time by assigning user jobs to nodes with the same control software version, and controls the execution order of update and user jobs accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If update jobs are executed on different nodes at different times, then node utilization efficiency is improved, but user jobs experience long delays and waiting times increase
Solution Approach 1:
The system performs preliminary scheduling of update jobs by identifying groups of nodes with identical control software versions and pre-planning which nodes will receive updates and in what sequence. This preliminary action allows the scheduler to minimize user job waiting times while maintaining high node utilization, as the scheduling decision is made in advance based on version groupings rather than ad-hoc execution
Solution Approach 2:
The patent applies local quality by treating nodes with identical control software versions as a homogeneous group for scheduling purposes. Within each version group, the system applies specific scheduling rules (updating all nodes in the group before or after user jobs), while different version groups may have different scheduling strategies. This localized approach optimizes the balance between update completion and user job execution for each group
2Loss of time
If update jobs are delayed to reduce user job waiting time, then user job execution is improved, but update job completion is postponed
Solution Approach 1:
The system dynamically adjusts the scheduling of update jobs based on real-time conditions. The scheduler evaluates user job requirements, node availability, and control software version distributions to determine the optimal timing for update execution. This dynamic approach allows the system to flexibly balance update completion timing with user job execution needs, rather than following a fixed update schedule
3Productivity
If nodes are assigned to user jobs without considering control software version, then node availability is maximized, but system reliability decreases due to version mismatches
Solution Approach 1:
The patent enforces homogeneity by requiring that all nodes assigned to execute a user job must have identical control software versions. The scheduler groups nodes by version and only assigns nodes from the same version group to the same user job. This ensures system reliability through version consistency while maintaining high node availability by utilizing all nodes across different version groups according to their availability status
Data Source
AI summary
A computer identifies an update job of updating a control software and two or more user jobs each with specification of the number of used nodes and an expected execution time. The computer plans a schedule of the update job and two or more user jobs so as to reduce the total waiting time of the two or more user jobs, on the basis of the number of used nodes and expected execution time in such a manner that each of the two or more user jobs is assigned to one or more nodes with the same version of the control software. The computer controls the order of execution of the update job and two or more user jobs according to the planned schedule.