Task Resource Scheduling via Priority Levels and Deadlines

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In traditional big data computations, tasks of high importance are delayed due to the first-come, first-compute rule, where low-importance tasks occupy resources, leading to impaired data presentation at the frontend.

Innovation Solution

A task resource scheduling method that determines priority levels and deadline times for tasks, sets a resource upper bound, and schedules tasks based on priority levels to avoid time resource conflicts, ensuring high-importance tasks are executed preferentially.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If tasks are executed according to a first-come, first-compute rule, then resource allocation is simple and fair, but high-importance tasks are delayed when low-importance tasks occupy resources

Engineering Contradiction:
Improveexecution efficiency of high-importance tasksVSAvoidscheduling complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent introduces priority levels as a new parameter to transform the scheduling system from a simple first-come-first-served approach to a priority-based scheduling mechanism. By assigning different priority levels to tasks based on their importance and deadline requirements, the system can differentiate resource allocation and execution order, thereby improving the execution efficiency of high-importance tasks without excessively increasing scheduling complexity.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If low-importance tasks are executed first, then resource utilization is maximized, but data presentation at frontend is impacted

Engineering Contradiction:
Improvedata presentation speedVSAvoidtask completion time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent performs preliminary classification and priority assignment to tasks before execution. By pre-determining the priority levels and expected completion times of tasks, the scheduling system can proactively allocate resources to high-importance tasks that require timely completion, preventing delays in data presentation at the frontend while still maintaining overall resource utilization.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If computing cluster resources are fully occupied, then resource utilization is high, but high-importance tasks cannot be computed on a timely basis

Engineering Contradiction:
Improvetimely completion of high-importance tasksVSAvoidresource availability
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent applies local quality by differentiating resource allocation based on task priority and characteristics. Instead of uniform resource distribution, the system allocates computing cluster resources selectively - providing preferential resource availability to high-importance tasks with tight deadlines while allowing lower-priority tasks to share remaining resources. This localized resource quality adjustment ensures timely completion of critical tasks without completely sacrificing overall resource utilization.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3438821B1Task-resource scheduling method and device
Publication Date: 2021.06.16 ALIBABA GROUP HOLDING LTD
  • EP3438821B1 patent drawingFigure 1
  • EP3438821B1 patent drawingFigure 2
  • EP3438821B1 patent drawingFigure 3

AI summary

The present invention provides a task resource scheduling method and apparatus, through determining priority levels of multiple tasks as well as set task deadline times, and determining a preset resource upper bound, based on task priority levels schedules the successive order of executing multiple tasks, and post-scheduling tasks satisfy a first condition, wherein, the first condition is that a time resource conflict does not exist for the tasks, the time resource conflict being that a section exists on a time series where more than one task overlaps, and the sum of the resources occupied in the section having the overlap is greater than the resource upper bound. When multiple tasks are executed according to the present example embodiment, the existing first-arriving, first-compute rule is no longer utilized, but rather scheduling is performed based on the priority levels of tasks; through the above-mentioned scheduling method, each task may be maintained as executed ordered by level, ensuring as much as possible that tasks corresponding to high-level requests preferentially obtain resources, and thereby may be preferentially executed, which is advantageous for the requests to present data at a frontend.