Virtual Machine Resource Optimization Through Load Forecasting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for optimizing virtual machine resources fail to account for future load conditions when relocating guest VMs, relying on human experience and time-consuming manual adjustments.
Innovation Solution
A resource optimization device and method that utilize time-series prediction algorithms to forecast future resource data for host virtual machines, determining high-load VMs and identifying suitable migration destinations based on historical and predicted data, considering periodicity and resource trends.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human optimization is used to relocate guest VMs, then migration decisions can be made based on experience and judgment, but the process is time-consuming and inefficient
Solution Approach 1:
The patent replaces manual human optimization with an automated prediction system that uses machine learning models to forecast VM load. The system automatically collects historical resource data, trains prediction models, and generates migration recommendations without human intervention, thereby eliminating the time-consuming nature of manual optimization while maintaining or improving decision accuracy through data-driven insights.
Solution Approach 2:
The system enables self-service optimization by automatically monitoring VM performance metrics, predicting future load conditions, and proposing migration targets without requiring human operators. The automated prediction model continuously learns from historical data and independently generates optimization recommendations, freeing operators from manual workload while ensuring consistent and scalable resource management.
2Productivity
If migration is based only on current load measurements, then immediate resource balancing can be achieved, but future load conditions and periodicity are not considered
Solution Approach 1:
The patent applies preliminary action by predicting future VM load conditions before migration decisions are made. The prediction model analyzes historical resource data to forecast upcoming load patterns, allowing the system to proactively identify migration opportunities before peak loads occur. This advance planning ensures that migrations are timed optimally to prevent future resource bottlenecks while maintaining current service levels.
Solution Approach 2:
The system implements feedback by continuously monitoring actual VM performance metrics and comparing them with predicted values. The prediction model is trained on historical data and continuously refined based on actual outcomes, creating a closed-loop system that improves accuracy over time. This feedback mechanism ensures that the system learns from past performance and adapts to changing workload patterns, maintaining high resource allocation efficiency.
3Adaptability or versatility
If manual optimization is performed, then migration decisions can be customized based on specific scenarios, but the process requires continuous human intervention and is not scalable
Solution Approach 1:
The patent achieves flexibility through parameter changes by allowing users to configure prediction model parameters, threshold values, and migration criteria according to specific organizational needs. The system can adjust prediction time horizons, resource threshold percentages, and migration triggers to match different workload patterns and service level requirements. This parameter-based customization maintains adaptability while the underlying automated prediction engine handles the complex decision-making processes.
Data Source
AI summary
A resource optimization device applies a time-series prediction algorithm to historical resource data for each of a plurality of host VMs to predict future predicted resource data for each of the plurality of host VMs; determines a high-load host VM among the plurality of VMs based on the historical resource data and the future resource data of each of the plurality of VMs; and determines a migration destination host VM to which a migration target guest VM of at least one of the high-load host VMs is to be moved.