Adaptable IO Performance Forecasting via Ensemble Model Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Storage systems face challenges in forecasting future resource consumption and preventing critical failures due to fluctuating demand, leading to inefficient resource allocation and potential system failures.
Innovation Solution
A computer-implemented method processes historical input/output (IO) performance data to train multiple IO modeling systems, determines forecast scores based on their performance, selects the most accurate models, and uses them to forecast future IO performance data, allowing for proactive resource allocation and alert generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple IO modeling systems are trained and evaluated, then forecasting accuracy is improved, but system complexity and computational resources increase
Solution Approach 1:
The system segments the forecasting task by training multiple specialized IO modeling systems, each optimized for different aspects of IO performance prediction. Each model is independently trained on specific portions of historical data and evaluated separately, allowing the system to manage complexity through modular architecture while improving overall forecasting accuracy through ensemble evaluation.
Solution Approach 2:
The system changes parameters such as training data portions, model hyperparameters, and evaluation metrics to optimize each IO modeling system's performance. By systematically varying these parameters across multiple models and using forecast scores based on median error values and standard deviations, the system achieves improved accuracy without proportionally increasing operational complexity.
2Reliability
If historical IO performance data is processed and models are retrained, then forecast reliability is improved, but processing time and computational cost increase
Solution Approach 1:
The system performs preliminary actions by pre-processing historical IO performance data, removing duplicate portions, and preparing training datasets before model training begins. This preliminary data preparation reduces processing time during subsequent model training and retraining operations, while maintaining forecast reliability through comprehensive data cleaning and validation.
Solution Approach 2:
The system implements feedback mechanisms by evaluating each IO modeling system's performance using forecast scores calculated from median error values and standard deviations. This feedback information is used to determine whether models should be retrained based on new IO performance data, allowing the system to maintain reliability while avoiding unnecessary retraining operations that would consume additional processing time.
3Measurement precision
If IO modeling systems are continuously monitored and updated, then forecast accuracy is maintained, but operational complexity increases
Solution Approach 1:
The system performs self-service by automatically monitoring its own forecast performance, calculating forecast scores for each IO modeling system, and determining when retraining is needed based on predefined criteria. This automated self-monitoring and self-evaluation mechanism maintains forecast accuracy while reducing operational complexity by eliminating the need for manual intervention in model performance assessment.
Solution Approach 2:
The system uses feedback from continuous performance monitoring to automatically adjust operations. Forecast scores derived from median error values and standard deviations provide feedback that triggers retraining only when necessary, maintaining forecast accuracy while simplifying operations through automated decision-making rather than manual monitoring and intervention.
Data Source
AI summary
A method, computer program product, and computing system for processing historical input/output (IO) performance data associated with one or more storage objects of a storage system. A plurality of IO modeling systems may be trained using the historical IO performance data. Modeling performance information may be determined for the plurality of IO modeling systems across the historical IO performance data. A forecast score may be determined for each IO modeling system based on the modeling performance information for the plurality of IO modeling systems. A subset of the plurality of IO modeling systems may be selected based upon the forecast score for each IO modeling system. The at least one IO modeling system may be trained using the historical IO performance data. IO performance data may be forecasted using the at least one trained IO modeling system from the subset of the plurality of IO modeling systems.