RAID Reconstruction via Disk Extent Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional RAID technologies face bottlenecks in data reconstruction due to limited I/O bandwidth of hot spare disks, particularly as storage capacity increases, making it difficult to reduce reconstruction time effectively.
Innovation Solution
The method involves determining RAID stripes for reconstruction based on I/O load states at a disk extent level, selecting RAID stripes with high or low I/O loads, and utilizing multiple storage processors to reconstruct data concurrently, optimizing the selection of spare disk extents to improve performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional RAID technology uses a single hot spare physical disk for reconstruction, then data redundancy and reliability are maintained, but the I/O bandwidth of the hot spare disk becomes a bottleneck, making it difficult to reduce reconstruction time
Solution Approach 1:
The patent segments the reconstruction process by dividing RAID stripes into multiple groups and assigning different groups to different storage processors for parallel reconstruction. This segmentation allows multiple hot spare disks to work simultaneously, breaking the single-disk bandwidth bottleneck while maintaining data redundancy through distributed parity reconstruction.
Solution Approach 2:
The patent merges multiple storage processors and hot spare disks into a coordinated reconstruction system. By combining the capabilities of multiple processors and disks working in parallel, the system achieves higher aggregate I/O bandwidth for reconstruction while preserving the reliability guarantees of traditional RAID through coordinated parity calculation and distribution.
2Device complexity
If traditional RAID technology relies on sequential reconstruction through a single storage processor, then system complexity is kept simple, but productivity during reconstruction is limited by the single processor's I/O bandwidth
Solution Approach 1:
The patent segments reconstruction tasks across multiple storage processors, with each processor handling specific RAID stripe groups. This segmentation increases reconstruction productivity by utilizing aggregate I/O bandwidth from multiple processors while managing complexity through structured task distribution and coordination mechanisms.
Solution Approach 2:
The patent introduces a new dimension of parallelism by utilizing multiple storage processors working simultaneously on different RAID stripes. This dimensional expansion from single-processor sequential execution to multi-processor parallel execution significantly boosts reconstruction productivity while complexity is managed through systematic task allocation.
Data Source
AI summary
Embodiments of the present disclosure provide a method and a device for managing storage system. In an embodiment, a plurality of RAID stripes to be reconstructed in the storage system are determined. A first RAID stripe is selected from a plurality of RAID stripes based on a plurality of I/O load states at a disk extent level for the plurality of RAID stripes at a first storage processor in the storage system, the first RAID having a high I/O load at the disk extent level. Then, a first disk extent for reconstructing the first RAID stripe is determined and the first storage processor is configured to reconstruct data for the first RAID stripe at the first disk extent. A corresponding device to implement the method is further disclosed.