Symbol-Level Declustering for RAID Data Corruption Isolation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current technologies, such as RAID 6 and T10 Data Integrity Field, face challenges in efficiently detecting and correcting silent data corruption in storage systems without significant reductions in usable storage capacity, especially when multiple sectors are corrupted.

Innovation Solution

The proposed solution involves symbol-level declustering of RAID stripes by dividing data into codewords with randomly distributed data and parity symbols, allowing for the identification of corrupt sectors using a probabilistic approach and bin counting algorithm, which requires only a single redundancy, thus being space-efficient.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If RAID 6 or T10 Data Integrity Field is used to detect and correct silent data corruption, then data reliability is improved, but usable storage capacity is significantly reduced

Engineering Contradiction:

Improvedata corruption detection and correction capabilityVSAvoidusable storage capacity

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments data into codewords with data symbols and parity symbols, randomly distributing symbols across drives. This segmentation approach enables corruption detection with only single redundancy per codeword, unlike RAID 6 which requires double parity across entire stripes, thereby preserving more usable storage capacity while maintaining reliability.

Inventive Principle:

Principle #1Segmentation

2Quantity of substance

If traditional RAID striping is used, then storage capacity is maximized, but the ability to detect and isolate corrupted sectors is insufficient

Engineering Contradiction:

Improvestorage capacityVSAvoidcorrupted sector identification capability

Core Design Contradiction:

Quantity of substanceVSDifficulty of detecting and measuring

Solution Approach 1:

The patent applies local quality by creating unique local characteristics for each codeword through random symbol distribution. Each codeword's symbols are distributed to specific drives, creating a distinctive pattern that enables identification of which sectors are corrupted when inconsistencies are detected, while maintaining high storage capacity utilization.

Inventive Principle:

Principle #3Local quality

3Reliability

If multiple parity schemes are implemented to handle multiple sector corruptions, then corruption detection capability is improved, but storage overhead increases

Engineering Contradiction:

Improvemultiple sector corruption handling capabilityVSAvoidstorage overhead

Core Design Contradiction:

ReliabilityVSLoss of substance

Solution Approach 1:

The patent employs dynamic probabilistic analysis instead of static multiple parity schemes. By calculating probabilities of corruption for each drive based on inconsistent codeword patterns, the system can handle multiple sector corruptions adaptively with only single redundancy, avoiding the fixed storage overhead of traditional multi-parity approaches.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10503597B2Symbol-level declustering of raid stripes to isolate data corruption

Publication Date: 2019.12.10 SEAGATE TECH LLC

AI summary

Data targeted for storage into a drive array is divided into codewords with data and parity symbols. The symbols of the codewords are randomly distributed across a stripe of the drive array. One or more drives affected by data corruption are found based on a probability that a subset of inconsistent codewords intersects the one or more drives.