RAID Storage Sets for Parallel Rebuild Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional RAID group repair methods are burdensome and can result in long rebuild times, increasing the risk of data loss due to bottlenecks in write performance, especially as the total capacity of individual physical drives grows, and mapped RAID spreads data across a large failure domain, complicating the rebuild process.
Innovation Solution
The method involves generating mutually exclusive sets of storage devices to form RAID stripes from respective storage extents, allowing for parallel data writing and limiting the failure domain, thereby improving rebuild efficiency and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional RAID repair methods are used to restore redundancy after disk drive failure, then data loss protection is maintained, but rebuild time increases and risk of data loss increases due to write performance bottlenecks
Solution Approach 1:
The patent segments the RAID group into multiple subsets, where each subset contains a portion of the data strips. When a failure occurs, only the affected subset needs to be rebuilt rather than the entire RAID group. This segmentation approach divides the large rebuild task into smaller, parallelizable units that can be reconstructed simultaneously from different source drives, significantly reducing overall rebuild time while maintaining data loss protection.
Solution Approach 2:
The patent introduces a new dimensional organization of RAID data by creating multiple subsets with overlapping data distribution across drives. Instead of the traditional single-dimension RAID stripe structure, the data is organized in a multi-dimensional manner where each subset can be independently reconstructed. This dimensional change enables parallel rebuild operations across multiple subsets simultaneously, addressing the time bottleneck without sacrificing reliability.
2Productivity
If mapped RAID techniques are used to distribute data across RAID extents, then rebuild speed improves through parallel writing, but failure domain expands across all drives in the system
Solution Approach 1:
The patent segments the failure domain by organizing drives into multiple subsets, where each subset is independently manageable. This segmentation limits the impact of failures to specific subsets rather than propagating across the entire RAID system. Combined with parallel rebuild capabilities within each subset, this approach achieves fast rebuild speeds while containing the effective failure domain to manageable portions of the system.
3Productivity
If RAID stripes are spread across all drives in the system, then data distribution is optimized, but rebuild complexity increases and risk of data loss during rebuild increases
Solution Approach 1:
The patent segments the RAID structure into multiple subsets, each containing a manageable portion of the data strips. This segmentation maintains efficient data distribution across drives while enabling independent management and reconstruction of each subset. When rebuilds are needed, only the affected subsets require attention, significantly reducing rebuild complexity compared to traditional approaches where the entire RAID group must be reconstructed.
Solution Approach 2:
The patent implements partial action by rebuilding only the necessary subsets rather than the entire RAID group. This partial rebuild approach reduces the scope and complexity of rebuild operations while maintaining overall data integrity. The system performs exactly the amount of rebuild work needed for the affected portions, avoiding the excessive action of rebuilding entire RAID groups even when only small portions are affected.
Data Source
AI summary
There is disclosed herein techniques for managing data storage. In one exemplary embodiment, the techniques comprise generating one or more sets of storage devices. Each set is configured to be mutually exclusive with respect to other sets of the one or more sets such that storage devices within the same set provide the basis for the formation of a RAID (Redundant Arrays of Independent Disks) stripe. The techniques further comprise forming a RAID stripe from respective storage extents associated with different storage devices of the same set. The techniques further comprise mapping a storage object to the RAID stripe. The techniques further comprise servicing host I/O (Input/Output) operations directed to the storage object by using the RAID stripe mapped thereto.