Storage Usage Prediction via Pattern Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting storage resource usage in cloud-based data storage services often result in significant discrepancies between predicted and actual usage, leading to unnecessary costs due to over-reservation of resources.
Innovation Solution
A method that classifies historical storage resource usage data into trend, cycle, and irregular patterns to accurately predict future usage by analyzing time series data, allowing for more precise estimation and cost management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to predict storage resource usage, then resource reservation is simplified, but prediction accuracy deteriorates leading to over-reservation and unnecessary costs
Solution Approach 1:
The patent segments the prediction process into distinct modules: data collection, pattern recognition (trend, cycle, irregular), and prediction generation. This segmentation allows each component to be optimized independently while maintaining overall system simplicity and improving accuracy through specialized processing at each stage.
Solution Approach 2:
The patent implements dynamic prediction by continuously monitoring actual usage against predicted values and adjusting future predictions based on deviations. The system adapts to changing usage patterns over time, transitioning from static reservation models to dynamic adjustment models that respond to real-time data, thereby improving accuracy without requiring complex manual configuration.
2Reliability
If storage resources are over-reserved to ensure availability, then service reliability is improved, but cost increases due to unused capacity
Solution Approach 1:
The patent employs feedback mechanisms where actual storage usage data is continuously fed back into the prediction system. The system compares predicted usage with actual usage, identifies deviations, and adjusts future predictions accordingly. This feedback loop enables the system to maintain reliable service availability while avoiding over-reservation by learning from actual usage patterns and correcting prediction biases.
Solution Approach 2:
The patent changes the parameters used for resource reservation from fixed conservative estimates to dynamic predictions based on identified patterns (trend, cycle, irregular). By transforming reservation parameters from static safety margins to pattern-based projections, the system achieves both reliability and cost efficiency by reserving only the capacity actually needed based on historical and predicted usage behavior.
3Measurement precision
If detailed pattern analysis is performed on historical data, then prediction accuracy is improved, but data processing time increases
Solution Approach 1:
The patent applies partial action by focusing pattern recognition on the most significant components of historical data rather than analyzing every detail equally. The system identifies and processes only the essential pattern types (trend, cycle, irregular) that have the greatest impact on prediction accuracy, filtering out redundant or less informative data elements to reduce processing time while maintaining high estimation accuracy.
Data Source
AI summary
Storage resource usage amount(s) are determined. For instance, storage resource usage data in a historical period related to a user is acquired. The pattern information of the storage resource usage data is determined according to a time series of the storage resource usage data, the time series being a series of observed values of the storage resource usage data in the historical period. In addition, the storage resource usage amount for a target period of the user can be determined based on the pattern information and the storage resource usage data. The pattern information at least includes at least one of a trend pattern, a cycle pattern, or an irregular pattern. Beneficially, a storage resource usage amount of a user can be more accurately determined in a future period, thereby providing the user with valuable reference information.