Predictive Storage Management Using Trend and Implementation Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional virtualized Storage Area Network (SAN) usage analysis is inadequate in predicting usage trends due to abrupt changes, periodicity, and varying growth rates, leading to inefficient storage device management functions such as real-time capacity provisioning and data purging.
Innovation Solution
An Information Handling System (IHS) with a predictive storage management engine that retrieves storage device usage data, generates usage trend models using statistical time-series analysis, and combines these with machine-learning-based storage system implementation models to forecast future usage, enabling proactive management actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional statistical time-series models are used to analyze virtualized SAN usage, then the analysis process is simple, but the prediction accuracy of usage trends is poor
Solution Approach 1:
The patent combines multiple modeling approaches (statistical time-series models and machine-learning-based implementation models) into a unified predictive storage management system. This integration allows the system to leverage the strengths of both approaches: statistical models capture temporal patterns in usage data, while machine-learning models account for implementation-specific characteristics, thereby improving overall prediction accuracy without being limited by the weaknesses of either individual approach
Solution Approach 2:
The predictive model functions as a composite analytical framework that integrates different types of models (statistical and machine-learning) to create a more robust prediction system. Just as composite materials combine different substances to achieve superior properties, this composite modeling approach combines different analytical methods to achieve superior prediction accuracy for storage usage trends
2Adaptability or versatility
If statistical time-series models are used for SAN usage analysis, then the implementation is straightforward, but the system cannot handle abrupt trend changes, step functions, periodicity, and varying growth rates
Solution Approach 1:
The system employs dynamic modeling capabilities through machine-learning-based implementation models that can adapt to changing usage patterns. These models can dynamically adjust to abrupt trend changes, step functions, and periodic variations in storage usage, unlike static statistical models. The machine-learning component continuously learns from new data, allowing the system to maintain high adaptability as usage patterns evolve over time
Solution Approach 2:
The patent utilizes parameter changes in machine-learning models to accommodate varying growth rates and complex usage patterns. By adjusting model parameters based on learned patterns from historical data, the system can adapt to different growth scenarios, periodic behaviors, and abrupt changes without requiring a complete model restructuring, thus achieving high versatility while managing complexity
3Productivity
If conventional storage management is used, then the system operation is simple, but real-time storage capacity provisioning recommendations and data purging functions are deficient
Solution Approach 1:
The predictive storage management system performs preliminary actions by generating forward-looking recommendations for storage capacity provisioning and data purging based on predicted future usage. Instead of reacting to storage issues after they occur, the system anticipates future storage needs and constraints, enabling proactive management actions that improve storage utilization efficiency and prevent service disruptions before they happen
Solution Approach 2:
The system implements continuous feedback loops where prediction results inform management decisions, and the outcomes of those decisions are fed back into the prediction model for continuous improvement. This feedback mechanism enables the system to learn from actual storage usage patterns and management actions, progressively enhancing prediction accuracy and storage management efficiency over time while maintaining automated operation
Data Source
AI summary
A predictive storage management system includes a storage system having storage devices, and a predictive storage management device coupled to the storage system via a network. The predictive storage management device includes a statistical time-series storage device usage sub-engine that retrieves first storage device usage data from a first storage device in the storage system and uses it to generate a first storage device usage trend model. A machine-learning storage system usage sub-engine in the predictive storage management device retrieves storage system implementation information from the storage system and uses it to generate a storage system implementation model. A storage management sub-engine in the predictive storage management device analyzes the first storage device usage trend model and the storage system implementation model to predict future usage of the first storage device and, based on that predicted future usage, performs a management action associated with the first storage device.