CN-122018789-A - Writing management method of NAND type solid-state storage device

Abstract

A write management method of a NAND type solid state storage device dynamically generates a write inhibit condition and controls a write operation by establishing a write context record, monitoring a write result and performing responsibility attribution analysis. After the data writing is completed, the ECC correction event, the retention error and the unexpected overwriting or reading failure event of the NAND block are monitored during the background observation period. When the monitoring result exceeds the preset tolerance range, backtracking the context record, judging the failure possibility of the combination of the specific write strategy and the block state, and establishing a corresponding write-in prohibition condition. And calculating the cumulative failure times of a single write strategy or a single NAND block, establishing strategy or block level limit, and dynamically updating the strategy and the block limit according to the observation period so as to ensure that the SSD can still effectively reduce the occurrence rate of high risk events in different load and aging stages, thereby improving the data reliability and the service life of the NAND block.

Inventors

• WANG CHONGYU

Assignees

• 晋达半导体有限公司

Dates

Publication Date

20260512

Application Date

20251231

Claims (4)

1. 1. A write-in management method of a NAND type solid-state storage device is characterized by comprising the following steps: (A) A writing context establishing step: Before each time data is written into a plurality of NAND blocks, establishing a writing context record, wherein the writing context record comprises a writing strategy, a plurality of NAND block states and a data attribute set, the writing strategy refers to a writing mode and parameters selected by a controller, the plurality of NAND block states refer to historical writing/erasing times and error conditions of corresponding NAND blocks, the data attribute set refers to the requirements of data on reliability or survival time, and the writing context record corresponds to the plurality of NAND blocks; (B) And a writing result observation step: Monitoring the plurality of NAND blocks associated with the write context record during a background observation period after completion of data writing, wherein the background observation period is a settable time for collecting ECC correction conditions, retention errors and unexpected overwrite or read failure events of the plurality of NAND blocks; (C) Responsibility attribution analysis step: When any monitoring item exceeds a preset tolerance range, the system backtracks the writing context record to judge whether the combination of the specific writing strategy and the states of the plurality of NAND blocks is related to a failure event or not, and further evaluates the possibility of failure caused by the writing strategy according to the past writing and observation results; (D) A write inhibit condition generation step of: If the responsibility attribution analysis determines that the combination of the specific write strategy and the plurality of NAND block states causes two consecutive adverse effects in the past, or that the combination of the specific write strategy and the plurality of NAND block states causes the total number of adverse effects to exceed three in the past, establishing a corresponding write inhibit condition, wherein the write inhibit condition is that the write strategy is limited or inhibited from being executed again under the condition of conforming to the specific NAND block state, and (E) The write-in management execution step: And managing and controlling writing operation according to the established writing prohibition condition, when receiving a writing request, the system firstly checks whether the identification information in the writing strategy of the writing request accords with any existing writing prohibition condition, and if so, refuses the writing and replaces other writing strategies or writes different NAND blocks.
2. 2. The method of claim 1, wherein the system further calculates a cumulative failure count of each of the write strategies for all of the plurality of NAND blocks, and if the cumulative failure count exceeds a predetermined threshold, establishes a corresponding strategy prohibition condition to restrict or prohibit write operations of the write strategy for all of the blocks, wherein the cumulative failure count is a total number of ECC corrections, retention errors, or write failure events caused by a combination of the strategy and writing of any of the plurality of NAND blocks during a background observation period.
3. 3. The method of claim 2, wherein the system further calculates a cumulative failure count for each of the plurality of NAND blocks under all write strategies, and if the cumulative failure count exceeds a predetermined threshold, a corresponding block constraint is established to limit or reduce the write operation of the NAND block, wherein the cumulative failure count refers to a total number of ECC corrections, retention errors, or write failure events caused by any combination of the write strategy and the plurality of NAND blocks during a background observation period.
4. 4. The write management method of claim 3, wherein the system automatically recalculates the number of combined failures of the write strategy and the plurality of NAND blocks and the number of accumulated failures of a single strategy or a single block during an observation period in each preset observation period, and dynamically updates the strategy prohibition condition and the block restriction condition according to the calculation result so as to ensure that the write management operation can continuously reduce the occurrence probability of high-risk combination and high-risk elements in different load and block aging stages.

Description

Writing management method of NAND type solid-state storage device Technical Field The invention relates to a control technology of a solid State Storage Device (SSD), in particular to a write management method for a NAND solid state storage device, which combines write strategy management, block health monitoring and dynamic prohibition condition generation to improve data reliability, reduce high risk write combination occurrence rate and prolong the service life of a NAND block. Background The conventional NAND-type solid-state storage device mostly adopts write optimization mechanisms such as data recovery, block balancing, cache management, etc., so as to maintain storage efficiency and prolong the service life of NAND blocks. These methods typically avoid excessive use of portions of the block by periodically moving the data or evenly distributing the write operations, thereby delaying block aging and improving the average lifetime. However, these techniques are mostly static or periodic, lack the ability to sense the health status of the blocks in real time, and lack the mechanism of continuous monitoring and dynamic adjustment for the data writing consequences, so that part of the high risk blocks may still frequently fail under the conditions of long-term use or high frequency writing. In practical applications, enterprise-level or high-performance storage systems often face large-scale, high-frequency data writing, and part of NAND blocks are prone to cumulative aging, resulting in reduced data retention, increased error correction times, and even unexpected write failures or data loss events. The prior art has been improved only for overall performance or average lifetime, lacking sophisticated management means for individual blocks or single write operations. Especially in a long-term operation or high-load environment, the health state of the blocks of the SSD may change rapidly, but the existing method cannot update the write strategy in real time or limit the high-risk operation automatically, which easily causes the decrease of the data reliability. In addition, the prior art relies on static rules or empirical parameters to perform write scheduling, such as preset block balancing or periodic data recovery, the adjustment frequency of which is independent of the health status of the blocks, and it is difficult to quickly respond to bursty or cumulative problems. For high frequency thermal data writing, a large number of writing operations may be carried out on a part of the blocks in a short time, so that the error accumulation speed exceeds the expected speed of the controller, and the health degradation in the short time is difficult to detect and process in real time in the prior art, thus increasing the risk of data loss. The prior art also lacks a unified failure event evaluation mechanism, and most of the prior art only performs average management for a single block or a single write strategy, cannot integrate historical write data, error events and block health status for full disc evaluation, and lacks means for dynamically generating write restriction conditions or adjusting write operations. Therefore, even if the SSD is configured with a health monitoring or error correction mechanism, high risk write combinations may occur in different operating conditions and aging stages due to lack of sophisticated, dynamic management, thereby causing data reliability degradation or block early failure. Therefore, it is necessary to provide a method for integrating write strategy determination, block health analysis and dynamic management control, so as to effectively reduce the occurrence rate of high risk events, improve the data reliability and prolong the service life of the NAND block. Disclosure of Invention An objective of the present invention is to provide a write management method for a NAND solid-state storage device, which dynamically generates write inhibit conditions and manages write operations by establishing write context records, monitoring write results and performing responsibility attribution analysis, so as to improve data reliability, reduce high risk write combination occurrence rate, and prolong the service life of each NAND block. The method can carry out refined judgment according to the data writing strategy, the block health state and the data attribute, and dynamically adjust the writing control strategy by combining background observation and accumulated failure analysis, so that the storage device can still maintain stable operation and data integrity in different operation loads and block aging stages. The invention provides a write management method of a NAND solid state storage device, comprising (A) a write context establishing step of establishing a write context record before each data write into a plurality of NAND blocks, wherein the write context record comprises a write strategy, a plurality of NAND block states and a data attribute set, wherein the wri