Dynamic Segment Ordering for Data Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional data recovery methods in memory subsystems are inefficient due to random or suboptimal ordering of recovery attempts for multiple translation units (TUs) with decoding failures, leading to decreased recovery chances and increased recovery time.
Innovation Solution
Implementing dynamic segment ordering by ranking TUs based on recovery likelihood metrics such as raw bit error rate (RBER) and voltage threshold distribution, prioritizing recovery attempts for TUs with the highest likelihood of success and abandoning those less likely to recover, thereby optimizing recovery time and increasing the chances of successful recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional random or suboptimal ordering methods are used for recovery attempts, then the data recovery process is simple to implement, but the recovery time increases and recovery chances decrease
Solution Approach 1:
The patent applies dynamics by transitioning from a static random ordering approach to a dynamic ordering approach where the recovery sequence is continuously adapted based on real-time metrics. The system dynamically adjusts the recovery order by monitoring RBER and voltage threshold distribution, selecting the next TU to recover based on current conditions rather than following a fixed predetermined sequence, thereby optimizing both recovery chances and time efficiency
2Productivity
If dynamic segment ordering based on RBER and voltage threshold distribution is implemented, then recovery efficiency and success rate improve, but the complexity of the data recovery process increases
Solution Approach 1:
The patent implements feedback mechanisms by continuously monitoring recovery metrics (RBER and voltage threshold distribution) and using this information to adjust the recovery ordering. The system feeds back the performance data from each recovery attempt to refine the ordering strategy for subsequent attempts, creating a closed-loop system that adapts to actual recovery conditions and improves efficiency while managing complexity through intelligent control
3Reliability
If all TUs with decoding failures are attempted to be recovered without prioritization, then complete recovery coverage is achieved, but unnecessary recovery attempts increase time consumption
Solution Approach 1:
The patent applies local quality by differentiating the treatment of different TUs based on their individual recovery likelihood characteristics. Instead of applying uniform recovery attempts to all failed TUs, the system identifies and prioritizes TUs with higher recovery likelihood (lower RBER, better voltage distribution) while deprioritizing or skipping those with lower likelihood, thereby allocating recovery resources more efficiently and reducing time spent on unlikely candidates
Data Source
AI summary
Methods, systems, and apparatuses include generating recovery likelihood metrics for undecodable segments in a stripe of data distributed across a redundant array of storage nodes. The recovery likelihood metrics are based on a determination of a likelihood of recovering the undecodable segment. The undecodable segments are ranked based on the recovery likelihood metrics. The undecodable segments are recovered in an order based on the ranking starting with the undecodable segment with the highest likelihood of recovery.