Disk Extent Reallocation via Wear-Aware Neighbor Matrix
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing disk reallocation methods in RAID systems face challenges in balancing disk extents evenly, leading to inefficiencies in load and wear level distribution, resulting in ping-pang reallocation between shuffling and load balancing operations.
Innovation Solution
An enhanced neighbor matrix is introduced, weighing disk wear levels to optimize resource reallocation, considering both distribution evenness and wear levels, and prioritizing disk pairs based on load levels to reduce ping-pang operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If disk extents are allocated evenly to all disks to improve rebuilding performance and parallel I/O processing, then disk extent distribution evenness is improved, but load and wear level balancing becomes difficult to maintain
Solution Approach 1:
The patent modifies the traditional neighborhood matrix by introducing wear level as a new parameter. Instead of only considering distribution evenness, the enhanced neighborhood matrix incorporates both distribution evenness and wear level information, allowing the system to make more informed reallocation decisions that balance both criteria simultaneously
Solution Approach 2:
The system continuously monitors disk wear levels and distribution characteristics, using this feedback to dynamically adjust reallocation strategies. The enhanced neighborhood matrix serves as a feedback mechanism that guides subsequent disk extent reallocation operations to maintain both evenness and wear balancing
2Stability of the object's composition
If shuffling operation is performed to improve disk extent distribution evenness, then distribution evenness is improved, but ping-pang reallocation between shuffling and load balancing operations occurs
Solution Approach 1:
The patent merges the shuffling operation and load balancing operation into a unified reallocation process guided by the enhanced neighborhood matrix. By combining these operations and using a single comprehensive evaluation criterion (enhanced neighborhood matrix), the system eliminates the ping-pang effect where operations repeatedly undo each other's progress
Solution Approach 2:
The enhanced neighborhood matrix is calculated in advance to evaluate both distribution evenness and wear level before reallocation occurs. This preliminary evaluation allows the system to plan reallocation operations that simultaneously improve both criteria, preventing subsequent ping-pang reallocation
3Reliability
If fine-grained load balancing is performed to balance IO load and wear level, then load and wear level balancing is improved, but disk extent distribution evenness deteriorates
Solution Approach 1:
The patent changes the evaluation parameters by introducing the enhanced neighborhood matrix that incorporates both distribution evenness and wear level. This allows the load balancing operation to consider distribution evenness as a constraint while optimizing for wear level balancing, preventing the deterioration of distribution evenness
Data Source
AI summary
Techniques perform resource reallocation for a disk system. Such techniques involve: determining, based on conditions of allocated disk extents in a plurality of disks and wear levels of the plurality of disks, an enhanced neighbor matrix characterizing both distribution evenness of the disk extents on the plurality of disks and the wear levels of the plurality of disks; and performing a resource reallocation operation on the plurality of disks based on the enhanced neighbor matrix. Accordingly, it is possible to, through the enhanced neighbor matrix that is based on the wear level, simultaneously consider both the distribution evenness of disk extents and the wear level of the disk while performing resource reallocation, thereby avoiding the problem of ping-pang resource reallocation caused by inconsistent standards between different resource reallocation methods, and having good compatibility with existing storage systems.