Data management method, storage device, and electronic device
By dynamically optimizing the management strategy of solid-state drives using reinforcement learning models, the performance and lifespan issues caused by fixed threshold management in existing technologies are resolved, enabling more refined storage management, improving storage performance and extending lifespan.
CN122152223APending Publication Date: 2026-06-05SLICONGO MICROELECTRONICS INC
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SLICONGO MICROELECTRONICS INC
- Filing Date
- 2026-01-30
- Publication Date
- 2026-06-05
AI Technical Summary
Technical Problem
Existing solid-state drive (SSD) garbage collection strategies are based on fixed thresholds or preset rules, which cannot achieve fine-grained management, resulting in insufficient storage performance and lifespan.
Method used
The reinforcement learning model is used to dynamically output optimized management decisions based on I/O command data and storage state data, including garbage collection strategies, wear leveling and data allocation, to achieve refined management.
Benefits of technology
It improves storage performance, extends the lifespan of storage devices, and meets data read and write needs in different scenarios.
β¦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122152223A_ABST
Abstract
Embodiments of the present application relate to the technical field of storage, and in particular to a data management method, a storage device and an electronic device. The method comprises: obtaining I / O command data, obtaining a first state vector based on the I / O command data; obtaining storage state data of the storage device; inputting the first state vector and the storage state data into a first reinforcement learning model to obtain a predicted management decision; updating storage management parameters of the storage device based on the predicted management decision; and performing a management operation on the storage device based on the storage management parameters. Embodiments of the present application actively output an optimized management decision from the feature data of the I / O command data and the storage state data of the storage device by the reinforcement learning model, realizing a leap from a traditional technology of "passive response to a fixed threshold" to "active perception-decision". Moreover, the storage management strategy can be specified as needed, and the data allocation logic can be optimized.
Need to check novelty before this filing date? Find Prior Art