Dynamic Computing Resource Scheduling via ML Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing resource scheduling methods are planning-intensive and less flexible, failing to efficiently manage resources by matching them to predefined schedules, leading to suboptimal utilization and increased costs.
Innovation Solution
A machine learning-based framework that clusters computing resources based on usage patterns and tags, forecasts idle times, and recommends dynamic scheduling to optimize resource allocation, allowing for more efficient utilization and cost savings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If predefined scheduling methods are used to manage computing resources, then resource allocation follows a fixed plan, but resource utilization efficiency deteriorates and costs increase
Solution Approach 1:
The patent transitions from static predefined schedules to dynamic machine learning-based scheduling recommendations. The system continuously learns from historical usage patterns and adapts scheduling recommendations in real-time, allowing resource allocation to dynamically respond to changing workload demands rather than following rigid predetermined plans.
Solution Approach 2:
The machine learning model enables the system to automatically analyze usage patterns and generate scheduling recommendations without requiring manual planning intervention. The system self-optimizes by learning from historical data and autonomously producing allocation recommendations, reducing reliance on predefined schedules created through manual planning processes.
2Ease of operation
If predefined schedules are used for resource allocation, then planning intensity is high, but flexibility deteriorates
Solution Approach 1:
The patent replaces manual planning mechanisms with machine learning-based automated scheduling. Instead of relying on human planners to create predefined schedules, the system uses ML models to automatically analyze usage patterns and generate flexible scheduling recommendations that adapt to changing conditions without requiring intensive manual planning efforts.
Solution Approach 2:
The system changes the fundamental parameter of scheduling from fixed predefined time slots to dynamic recommendations based on learned usage patterns. The ML model continuously adjusts scheduling parameters based on historical data analysis, enabling the system to flexibly adapt to varying workload demands rather than being constrained by static schedule parameters.
3Reliability
If computing resources are kept running to ensure availability, then service reliability is maintained, but energy consumption and costs increase
Solution Approach 1:
The machine learning model performs preliminary analysis of historical usage patterns to predict future resource needs. By learning from past data, the system can anticipate when resources will be needed and schedule their activation in advance, ensuring availability when required while avoiding continuous operation during predicted low-demand periods, thus reducing energy consumption.
Solution Approach 2:
The system implements feedback loops where usage patterns are continuously monitored and fed back to the ML model. This feedback mechanism allows the system to learn from actual resource utilization and adjust scheduling recommendations accordingly, optimizing the balance between maintaining service availability and reducing energy consumption based on real-world performance data.
Data Source
AI summary
Properties associated with computing resources are received. At least a portion of the received properties is used to cluster the computing resources into one or more operating groups. At least a portion of the received properties is used to determine a recommendation of an operation schedule for at least one of the one or more operating groups. The recommendation is provided. A feedback is received in response to the recommendation.