Hybrid RAID-DSN Storage for Data Reliability

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current cloud storage systems face challenges in efficiently writing and reading data due to limitations in error correction schemes, particularly in dispersed storage networks, which can lead to data loss and security vulnerabilities.

Innovation Solution

A dispersed storage network (DSN) architecture that utilizes error encoding and decoding techniques, such as Cauchy Reed-Solomon encoding, to distribute data across multiple storage units, ensuring data integrity and security through redundant encoding and decoding processes, allowing for efficient storage and retrieval without the need for redundant copies.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data is stored using traditional RAID systems or simple dispersed storage, then storage capacity and basic redundancy are provided, but data security against unauthorized access and robustness against multiple failures are insufficient

Engineering Contradiction:
Improvedata security and robustnessVSAvoidencoding and decoding process complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments data into multiple data slices and applies different encoding strategies to different segments. Specifically, it divides data into hot data (frequently accessed) and cold data (infrequently accessed), then applies distinct encoding schemes: Cauchy Reed-Solomon encoding for hot data and simpler parity encoding for cold data. This segmentation allows the system to achieve high security for critical data while maintaining efficiency for less critical data, resolving the contradiction between security and complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the encoding parameters based on data characteristics and access patterns. For hot data requiring high security, it uses Cauchy Reed-Solomon encoding with specific threshold parameters (k data slices needed to reconstruct). For cold data, it uses simpler encoding with different parameters. This dynamic parameter adjustment allows the system to optimize both security and computational complexity based on actual needs.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If Cauchy Reed-Solomon encoding is applied to all data, then high data security and fault tolerance are achieved, but computational overhead and storage efficiency decrease

Engineering Contradiction:
Improvedata integrity and securityVSAvoidwriting and reading efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent applies the principle of local quality by using different encoding schemes for different data portions based on their access patterns and security requirements. Hot data (frequently accessed) receives Cauchy Reed-Solomon encoding for high security, while cold data (infrequently accessed) uses simpler encoding. This localized application of quality measures ensures high productivity for cold data while maintaining high reliability for hot data, resolving the contradiction between security and efficiency.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent applies partial action by not using the most robust encoding (Cauchy Reed-Solomon) for all data, but only for the portion that requires it (hot data). For cold data, it uses minimal sufficient encoding. This partial application of excessive security measures where needed and reduced measures where sufficient resolves the contradiction between achieving high reliability and maintaining productivity.

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If multiple redundant copies of data are stored, then data availability and security are improved, but storage space consumption increases

Engineering Contradiction:
Improvedata availabilityVSAvoidstorage space
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments data into slices and distributes them across multiple storage units with error encoding, rather than creating full redundant copies. The segmentation allows the system to achieve data availability through distributed storage with fewer total copies, as error encoding enables reconstruction from a threshold number of slices rather than requiring complete copies.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the redundancy parameter from traditional RAID approaches (fixed redundancy levels) to threshold-based redundancy where only k out of n slices are needed to reconstruct data. This parameter change allows more efficient use of storage space while maintaining data availability, as the system can tolerate more failures with the same storage capacity.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10146645B2Multiple memory format storage in a storage network
Publication Date: 2018.12.04 PURE STORAGE INC
  • US10146645B2 patent drawing
  • US10146645B2 patent drawing
  • US10146645B2 patent drawing

AI summary

A method includes sending a data retrieval request regarding a data segment of a data object to redundant array of independent disk (RAID) memory and to dispersed storage network (DSN) memory. The method further includes receiving a first read response from a first one of the RAID memory and the DSN memory. The method further comprises recovering the data segment from the first read response. The method further includes determining whether to wait for a second read response from a second one of the RAID memory and the DSN memory. When the computing device determines to wait, the method further includes receiving the second response within a given time frame. The method further includes recovering a copy of the data segment from the second read response. When the data segment substantially matches the copy of the data segment, the method further includes utilizing either the data segment or the copy of the data segment.