Distributed Allocation Device for Computing Process Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for allocating computing processes to multiple nodes in factory automation do not adequately consider precedence constraints, leading to inefficient deadline setting and potential delays in data analysis, especially when deadlines exceed data collection periods.
Innovation Solution
A distributed allocation system that uses a device connected to multiple computing nodes to input network architecture, process names, execution periods, deadlines, precedence constraints, and data communication volumes, creating a constraint condition equation to determine the order of process execution and allocate tasks efficiently, ensuring deadlines are met while avoiding unnecessary resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If computing processes are distributed to multiple computing nodes, then the processing load and access load on the server are reduced, but the complexity of allocating processes while satisfying deadlines increases
Solution Approach 1:
The patent segments the computing processes into groups based on precedence constraints, where each group contains processes that must execute in a specific order. This segmentation allows the allocation algorithm to handle constrained processes as unified units, reducing the complexity of satisfying deadline requirements while distributing processes across multiple computing nodes.
Solution Approach 2:
The patent performs preliminary classification of computing processes into groups according to their precedence constraints before allocation. By pre-organizing processes into executable units that respect dependency relationships, the system simplifies the subsequent allocation and scheduling operations, making it easier to satisfy deadlines in distributed environments.
2Loss of time
If deadlines are set for each computing process, then the required time from start to end can be controlled, but the difficulty of setting appropriate deadlines considering precedence constraints and communication volumes increases
Solution Approach 1:
The patent calculates and sets deadlines for computing process groups in advance, considering the cumulative execution time and communication volumes within each group. By performing deadline setting as a preliminary step before allocation, the system simplifies the complexity of coordinating multiple individual process deadlines while respecting precedence constraints.
Solution Approach 2:
The patent merges multiple computing processes into groups based on precedence constraints, and sets a unified deadline for each group rather than individual deadlines for every process. This merging approach reduces the number of deadline parameters to manage and simplifies the overall deadline control mechanism in distributed systems.
3Reliability
If the deadline is longer than the data collection period, then the computing process can be executed more thoroughly, but the resource allocation becomes inefficient with unnecessary allocations
Solution Approach 1:
The patent applies partial action by allocating computing processes only during the necessary time window within the deadline, rather than continuously occupying resources. When the actual execution time is less than the deadline, the excess time is used for other purposes, achieving efficient resource utilization while still completing thorough analysis when needed.
Solution Approach 2:
The patent implements dynamic resource allocation where computing nodes can be reassigned for different tasks after completing their assigned computing processes, even if the deadline has not been reached. This dynamic approach allows resources to be flexibly reallocated based on actual execution progress, improving overall system productivity while maintaining reliability.
Data Source
AI summary
Computing processes are allocated to a plurality of computing nodes while taking a precedence constraint among a plurality of computing processes into consideration even when a deadline is longer than a period of the computing process. In a distributed allocation device, a creation unit determines an order to execute the computing processes based on the precedence constraint, and creates a constraint condition equation satisfying the deadline based on an execution time of the computing processes and a data communication volume among the computing processes. If the deadline is longer than the entire period, the creation unit creates the constraint condition equation which does not allocate an other computing process during execution of an excess computing process, in the subsequent entire period, to the computing node to which the excess computing process is to be allocated, which is executed past the entire period.