Database Storage Forecasting with ML to Prevent Abends
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing database management systems face challenges in accurately forecasting storage requirements, leading to potential database failures due to insufficient storage, and inefficient use of storage resources due to overallocation.
Innovation Solution
A computing system utilizing machine learning models to retrieve database state data, determine average and peak storage usage thresholds, and forecast additional storage needs to prevent abnormal database termination, while also deallocating excess storage for more efficient resource use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If storage is allocated to prevent database failures, then reliability improves, but storage resources may be excessively used
Solution Approach 1:
The system performs preliminary storage requirement forecasting using machine learning models before database failures occur. By predicting future storage needs based on historical data patterns, the system proactively allocates storage resources to prevent failures while avoiding overallocation through data-driven accuracy.
Solution Approach 2:
The system continuously monitors actual database storage usage and compares it against forecasted requirements. This feedback mechanism allows the system to adjust storage allocations dynamically, ensuring sufficient resources are allocated to prevent failures while deallocating excess storage to optimize resource utilization.
2Quantity of substance
If manual storage management is performed, then storage optimization is achieved, but computational resources and time are excessively consumed
Solution Approach 1:
The system implements self-service storage management where the machine learning model automatically forecasts storage requirements and the system autonomously performs allocation and deallocation decisions. This eliminates the need for manual intervention while optimizing storage resource efficiency through continuous automated monitoring and adjustment.
3Reliability
If storage is overallocated to databases, then database failures are prevented, but computational resources are wasted
Solution Approach 1:
The system dynamically adjusts storage allocation parameters based on forecasted database requirements rather than using fixed overallocated thresholds. By continuously updating allocation decisions based on actual usage patterns and predictions, the system maintains reliability while minimizing computational resource waste through precise, data-driven parameter adjustment.
Data Source
AI summary
Aspects of the disclosure relate to using machine learning models to automatically forecast amounts of storage to allocate to databases. A computing system may retrieve database state data that indicates states of databases. Based on inputting the database state data into machine learning models, average storage usage thresholds and peak storage usage thresholds may be determined for the databases. The average storage thresholds may indicate an average amount of storage used by the databases. The peak storage usage thresholds may indicate an amount of storage that is allocatable without causing an abnormal termination. Based on inputting the database state data, the average storage usage threshold, and the peak storage usage threshold into the machine learning models, a forecasted amount of additional storage to allocate to the databases to prevent abnormal termination of the database may be determined. Further, the additional storage may be allocated based on configurations of each database.