Storage System Parity Overhead Reduction via Chunk Migration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing storage systems face inefficiencies in managing storage areas within RAID groups, leading to excessive parity calculation overhead and suboptimal utilization of storage space, which deteriorates response performance when handling I/O requests from host computers.
Innovation Solution
The storage apparatus releases underutilized management units and migrates allocated pages to other management units, allowing for more equitable page allocation across virtual volumes and improving storage area utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple pages belonging to the same management unit are allocated to different virtual volumes, then storage area utilization is improved, but parity calculation overhead becomes excessively large
Solution Approach 1:
The patent segments the management unit into smaller granular units (chunks) and introduces a chunk management table to track allocation status. This segmentation allows selective allocation of individual chunks to different virtual volumes while maintaining parity calculation efficiency by processing at the chunk level rather than the entire management unit level.
Solution Approach 2:
The patent applies local quality by differentiating the treatment of different chunks within the same management unit. Each chunk can be independently allocated to different virtual volumes based on local demand, while the parity calculation is optimized for each local chunk's specific allocation pattern, rather than uniformly treating the entire management unit.
2Adaptability or versatility
If pages are allocated equally to multiple pool volumes, then allocation fairness is improved, but response performance deteriorates due to unequal actual utilization
Solution Approach 1:
The patent introduces dynamic allocation mechanisms where the management device continuously monitors actual utilization patterns and adjusts chunk allocations accordingly. The allocation strategy transitions from static equal distribution to dynamic reallocation based on actual I/O patterns, allowing the system to adapt to changing workload conditions and optimize response performance while maintaining fairness.
Solution Approach 2:
The patent implements feedback mechanisms where utilization information from pool volumes is fed back to the management device. This feedback loop enables the management device to identify underutilized volumes and redistribute chunks to balance the allocation, ensuring both fairness and optimal response performance through continuous adjustment based on actual utilization metrics.
3Device complexity
If management units are created within RAID groups, then storage organization is improved, but page migration flexibility is reduced
Solution Approach 1:
The patent segments the management unit into smaller chunks, creating a hierarchical structure that maintains the organizational benefits of RAID groups while introducing flexibility at the chunk level. This segmentation allows pages to be migrated between different chunks within the same management unit, providing migration flexibility without disrupting the RAID group structure.
Solution Approach 2:
The patent adds a new dimension of organization by introducing chunks as an intermediate layer between RAID groups and individual pages. This dimensional addition creates multiple levels of organization (RAID group > management unit > chunk > page), enabling flexible page migration within the hierarchical structure while maintaining the benefits of RAID group organization.
Data Source
AI summary
This invention is intended for the purpose of providing the storage system, the storage apparatus, and the storage system by which, even if the storage areas allocated to the virtual volume are managed in management units set by the RAID group, overhead for parity calculation does not become excessive. This invention, by releasing a specific management unit not fully utilized for page allocation from allocation to the virtual volume and migrating the allocated pages belonging to this specific management unit to the other management unit, makes the storage areas of the specific management unit available for the write accesses for the other virtual volumes from the host computer.