Virtual Chunk Spare Allocation for RAID Rebuild Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional RAID systems face challenges in efficiently managing spare drives and rebuilding data due to the increased use of high-capacity, low-cost drives, which are less reliable, leading to longer rebuild times and increased maintenance efforts, especially when there is a lack of spare drives.
Innovation Solution
A computer system that forms virtual parity groups and physical parity groups, maps virtual stripes to different physical storage drives, and converts selected virtual chunks to spare regions, allowing for efficient management and allocation of spare regions without the need for additional spare drives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If high-capacity, low-cost drives (SATA/NL-SAS) are deployed to increase storage capacity and reduce costs, then storage capacity increases and costs decrease, but drive reliability decreases leading to longer rebuild times and increased maintenance efforts
Solution Approach 1:
The system segments the storage pool into virtual chunks of fixed size (e.g., 512 GB), allowing granular management of storage resources. This segmentation enables the creation of multiple virtual spares from a single physical drive failure scenario, improving reliability without requiring additional physical drives.
Solution Approach 2:
The patent introduces a virtualization layer (virtual storage drives, virtual parity groups, virtual chunks) as an intermediary between physical drives and logical volumes. This virtual layer abstracts the physical drive failures and enables rapid data reconstruction by mapping failed drive data to available virtual spares, resolving the reliability issue while maintaining high-capacity drive deployment.
2Quantity of substance
If high-capacity drives are used, then storage capacity increases, but rebuild time increases due to larger data volumes
Solution Approach 1:
By segmenting the storage space into fixed-size virtual chunks (e.g., 512 GB), the system enables parallel reconstruction of multiple virtual spares simultaneously. When a physical drive fails, multiple virtual chunks can be reconstructed in parallel across different physical drives, significantly reducing total rebuild time compared to sequential reconstruction of entire drive capacities.
Solution Approach 2:
The system reconstructs data in partial units (virtual chunks of fixed size) rather than waiting to reconstruct entire drive capacities. This allows the rebuild process to proceed in manageable increments, reducing the time penalty associated with high-capacity drives while maintaining data integrity.
3Reliability
If conventional RAID techniques are used to manage spare drives, then fault tolerance is maintained, but the process overhead for modifying RAID levels is high and time is required to attain spare regions
Solution Approach 1:
The system performs preliminary actions by pre-allocating fixed-size virtual chunks that can serve as virtual spares. When a drive failure occurs, the virtualization layer can immediately designate available virtual chunks as spares without requiring RAID level modifications or complex reconfiguration, eliminating the time penalty associated with conventional RAID spare allocation.
Solution Approach 2:
The patent implements a dynamic virtualization layer that can flexibly allocate and reallocate virtual chunks as spares based on current system state. This dynamic approach allows rapid response to drive failures by programmatically creating virtual spares from available capacity, avoiding the static and time-consuming RAID level modification processes of conventional systems.
Data Source
AI summary
A virtual parity group including virtual storage drives is formed. A redundant data set is stored in a virtual stripe array including virtual stripes that are virtual storage regions of a prescribed size of the respective virtual storage drives constituting the virtual parity group. The virtual stripes of the virtual stripe array are mapped to different physical storage drives, respectively. Virtual chunks including one or more virtual stripe arrays formed in the virtual parity group. Physical parity groups including the physical storage drives are formed. The virtual stripes of the virtual chunks in a distributed manner are mapped to the physical parity groups. One or more virtual chunks are selected from among the virtual chunks. Storage regions corresponding to the selected one or more virtual chunks are converted to spare regions in the physical storage region.