Resource Management Platform Task Allocation for AI Data Sets

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing resource management platforms waste time and resources due to the large amount of data sets needed for artificial intelligence model training and testing, as they allocate tasks based solely on node performance without considering data set availability and size, leading to inefficient task processing.

Innovation Solution

A task allocation method and system that acquires data set distribution information, scores nodes based on data set size, and selects nodes with the required data sets meeting preset requirements for task allocation, reducing unnecessary data downloads and network pressure.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If tasks are allocated based on node performance without considering data set availability, then task allocation speed is improved, but data download time and network pressure increase significantly

Engineering Contradiction:
Improvetask allocation speedVSAvoiddata download time
Core Design Contradiction:
SpeedVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-distributing data sets to nodes before task allocation occurs. The data set management module proactively copies required data sets from storage to multiple nodes in advance, so when tasks are allocated, nodes already have the necessary data locally, eliminating download delays during task execution

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

A data set management module acts as an intermediary between storage and nodes. This module handles data set distribution, caching, and management, mediating between the need for data availability and the desire for fast task allocation by maintaining local copies of data sets on nodes

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If data sets are downloaded to each node before task processing, then task processing reliability is improved, but resource consumption and time cost increase

Engineering Contradiction:
Improvetask processing reliabilityVSAvoidresource consumption
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The system applies local quality by distributing data sets selectively to specific nodes based on their assigned tasks rather than universally to all nodes. Each node receives and stores only the data sets required for its specific task, creating local data availability without redundant duplication across the entire system

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system changes the parameter of data location from centralized storage to distributed local storage on nodes. By transforming where data resides (from central storage to node-local cache), the system ensures reliable access for task processing while reducing the total amount of data that needs to be actively managed and transferred

Inventive Principle:
Principle #35Parameter changes

3Productivity

If data sets are pre-distributed to multiple nodes, then task processing efficiency is improved, but system complexity increases

Engineering Contradiction:
Improvetask processing efficiencyVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The data set management module performs multiple functions: it manages data set storage, handles distribution to nodes, maintains local caches, and coordinates with the task allocation module. This multi-functional approach consolidates complexity into a single management entity rather than requiring separate mechanisms for each function

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system implements feedback mechanisms where nodes report their data set status and task completion status back to the data set management module. This feedback allows the system to dynamically adjust data distribution, track which nodes have which data sets, and optimize future allocations based on actual system state

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4020200B1Resource management platform-based task allocation method and system
Publication Date: 2024.07.24 GUANGDONG INSPUR BIG DATA RES CO LTD
  • EP4020200B1 patent drawingFigure 1
  • EP4020200B1 patent drawingFigure 2
  • EP4020200B1 patent drawingFigure 3

AI summary

The present application discloses a task allocation method and system based on a resource management platform. The method comprises: receiving an artificial intelligence model training and/or testing task and a name of data set required for processing the task; acquiring data set distribution information of a plurality of nodes; judging if the node has the required data sets according to names of the data sets in the node; and selecting a node with the size of the required data set meeting preset requirements for task allocation according to the size of the required data set in the node if the node has the required data set. It may be seen that, in the present application, the situation of data sets in a node is taken into account during task allocation, and the node with the size of the required data set meeting preset conditions is selected for task allocation, such that the node does not need to download the required data set or reduce the frequency of downloading data sets by a node, thereby improving the efficiency in processing tasks.