Job Scheduling via Modelled Skyline Alignment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Scheduling recurring computing jobs in distributed systems faces challenges in effectively utilizing resources, balancing the risk of overloading nodes while avoiding under-utilization, and adhering to service-level agreements (SLAs) that require timely job completion without wasting resources.

Innovation Solution

A job scheduling system that acquires and normalizes time series metric values to form a metric-based skyline, uses best-fit analysis through linear programming with over-allocation and under-allocation penalties to align execution times, minimizing the risk of SLA violations and resource under-utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If jobs are scheduled based on current workload to prevent node overload, then system reliability is improved, but resource utilization deteriorates

Engineering Contradiction:

Improvesystem reliabilityVSAvoidresource utilization

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system performs preliminary actions by acquiring time series metric values and forming modelled skylines before scheduling jobs. It predicts future workload patterns using historical data and linear programming to identify optimal execution times in advance, preventing both overload and under-utilization before they occur

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback by continuously acquiring time series metric values during job execution and using this data to refine future scheduling decisions. The modelled skylines are updated based on observed workload patterns, creating a closed-loop system that adapts to actual system behavior

Inventive Principle:

Principle #23Feedback

2Productivity

If jobs are scheduled to maximize resource utilization, then productivity is improved, but the risk of node overload increases

Engineering Contradiction:

Improveresource utilizationVSAvoidsystem reliability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system changes parameters by transforming raw time series metric values into normalized modelled skylines that represent workload patterns. It uses linear programming to determine optimal scheduling parameters that balance utilization and reliability, dynamically adjusting execution times based on predicted workload conditions

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system performs preliminary analysis by forming modelled skylines and identifying optimal execution times before jobs are scheduled. This advance planning ensures resources are fully utilized while preventing overload conditions through predictive workload assessment

Inventive Principle:

Principle #10Preliminary action

3Reliability

If execution times are optimized to meet SLA requirements, then service level agreement compliance is improved, but resource waste increases

Engineering Contradiction:

ImproveSLA complianceVSAvoidresource waste

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The system transforms execution time parameters by aligning modelled skylines to identify optimal start times that meet SLA requirements. It changes the scheduling approach from fixed intervals to dynamically optimized times based on predicted workload patterns, ensuring SLA compliance while minimizing resource idle time

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system applies partial action by scheduling jobs at specific optimal times within available windows rather than continuously. This selective scheduling meets SLA requirements while avoiding excessive resource allocation during periods when workload conditions are unfavorable

Inventive Principle:

Principle #16Partial or excessive action

4Ease of operation

If simple scheduling rules are used to reduce complexity, then ease of operation is improved, but the ability to balance workload and utilization deteriorates

Engineering Contradiction:

Improvescheduling simplicityVSAvoidworkload balancing

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The system creates simplified representations (copies) of complex workload patterns through modelled skylines. These abstracted models capture essential workload characteristics without the full complexity of raw time series data, enabling effective scheduling decisions through streamlined analysis

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system transforms complex time series data into simplified modelled skyline parameters that retain essential workload information. This parameter transformation maintains workload balancing capability while reducing computational complexity for scheduling decisions

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10503548B2Resource and latency estimation-based scheduling in a distributed computing environment

Publication Date: 2019.12.10 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

A time series of metric values indicative of a cost of executing a job may be acquired. The acquired time series may be normalized to form a skyline indicative of costs incurred, over time, during an instance of executing the job. A modelled skyline may be formed by a best-fit analysis of a plurality of metric-based skylines, constrained by penalties for over-allocation and under-allocation. Based on the modelled skyline, the modelled skyline may be aligned with one or more additional modelled skylines to identify execution times for the job. The identified execution times may be selected to minimize risk of exceeding a time-to-complete parameter and avoid under-utilization of resources.