NAND Flash Data Integrity via Adaptive Patrol Rewriting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
NAND flash memory experiences data degradation due to wear, leading to incorrect data reproduction, especially when errors exceed the correction capabilities of error correction codes (ECCs).
Innovation Solution
A non-volatile semiconductor storage apparatus with a management module, sensor, decision module, error correction module, error detection module, and rewrite module that manages rewrite counts and temperature to determine patrol and rewrite periods, performing error correction and rewriting data to maintain data robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If error correction codes (ECCs) are used to correct data errors, then data reliability is improved, but the system cannot handle errors beyond the correction capabilities of ECCs
Solution Approach 1:
The patent implements patrol reads that periodically read and rewrite data before errors develop beyond correction capability. The management module monitors rewrite counts and temperature to determine optimal patrol intervals, performing preventive maintenance before failure occurs. This preliminary action ensures data remains correctable by ECC even as wear progresses.
Solution Approach 2:
The system continuously monitors data integrity through error detection during patrol reads and uses this feedback to adjust rewrite strategies. When errors are detected or wear thresholds are approached, the management module modifies patrol and rewrite schedules dynamically, optimizing the balance between data reliability and storage lifespan.
2Reliability
If patrol and rewrite operations are performed frequently to maintain data robustness, then data reliability is improved, but storage lifespan and productivity are reduced
Solution Approach 1:
The patent implements dynamic patrol and rewrite scheduling based on real-time monitoring of rewrite counts and temperature. The management module adjusts patrol intervals and rewrite frequencies according to actual wear conditions, transitioning from static fixed-interval operations to adaptive dynamic operations that optimize the reliability-lifespan tradeoff.
Solution Approach 2:
The system changes operational parameters (patrol interval, rewrite frequency) based on monitored conditions. As rewrite counts increase or temperature rises, the management module modifies these parameters to reduce unnecessary operations while maintaining sufficient data robustness, thereby extending storage lifespan.
3Reliability
If rewrite operations are performed to correct data errors, then data integrity is improved, but the time to retain stored data shortens due to increased wear
Solution Approach 1:
The patent performs preliminary patrol reads and selective rewrites before data integrity deteriorates beyond correction capability. By acting in advance on wear-induced degradation, the system maintains data integrity longer without requiring frequent corrective rewrites, thus extending the time to retain stored data.
Solution Approach 2:
The system creates copies of data through patrol reads and preserves them through controlled rewrites. This copying mechanism allows the original data to remain intact while maintaining integrity through redundant copies that can be restored if degradation occurs, reducing the frequency of disruptive rewrite operations.
Data Source
AI summary
According to one embodiment, a non-volatile semiconductor storage apparatus is configured to decide determination periods respectively corresponding to each of management blocks based on rewrite count information items and a temperature, and to perform a determination processing for each of management blocks for each determination period. The determination processing includes determining whether first data read from a block in the blocks is normal based on the number of errors that are occurred in the first data. The apparatus is configured to perform a rewrite processing of rewriting the first data to second data which is error-corrected when it is determined that the first data is not normal.