Virtual Node Migration for Heterogeneous Hard Disk Capacity Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Massive storage systems face challenges in supporting hard disk capacity heterogeneity, leading to reduced user experience due to existing storage solutions' inability to efficiently manage differing disk capacities, especially during expansion and replacement.
Innovation Solution
A data storage method that detects actual usage of hard disks, migrates virtual nodes based on capacity levels, and adjusts migration ratios to balance storage load across disks of varying capacities, ensuring efficient use of storage space and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If consistent hashing algorithm is used to distribute objects across physical nodes, then balance and dispersibility are improved, but absolute balance cannot be achieved when physical nodes are relatively few
Solution Approach 1:
The patent segments the storage system into virtual nodes and physical nodes, where each physical node is divided into multiple virtual nodes. This segmentation allows the system to achieve better load distribution by distributing objects across virtual nodes rather than directly to physical nodes, solving the problem of imbalance when physical nodes are few.
Solution Approach 2:
The patent introduces a virtual dimension layer between physical nodes and objects. By mapping objects to virtual nodes which then map to physical nodes, the system adds an intermediate dimension that enables more granular control over distribution, achieving absolute balance even with limited physical nodes.
2Manufacturing precision
If virtual nodes are allocated in synchronous growth to maintain balance, then storage capacity balance is improved, but the system cannot support hard disk capacity heterogeneity during expansion
Solution Approach 1:
The patent implements dynamic virtual node allocation where the number of virtual nodes per physical node is not fixed but adjusts based on actual storage capacity. When disks are added or replaced with different capacities, the system dynamically recalculates and reallocates virtual nodes to maintain balance, supporting capacity heterogeneity while preserving load distribution equilibrium.
Solution Approach 2:
The patent changes the parameter of virtual node quantity from a static value to a dynamic value that depends on disk capacity. By making the virtual node allocation parameter variable rather than constant, the system can adapt to different disk capacities and maintain balance across heterogeneous storage resources.
3Manufacturing precision
If same quantity of virtual nodes are allocated to each physical node, then initial balance is improved, but storage space cannot be adequately used when disk capacities differ
Solution Approach 1:
The patent applies local quality by allocating different numbers of virtual nodes to different physical nodes based on their individual capacities. Instead of uniform allocation, each node receives virtual nodes proportional to its storage capacity, ensuring that larger disks handle more load while smaller disks handle less, thereby achieving both balance and high utilization.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of the present invention disclose a data storage method and a storage server, where the method includes: detecting actual usage of a hard disk corresponding to an initial capacity level in configured hard disk capacity levels, where the hard disk capacity levels are obtained by means of classification according to an actual hard disk capacity; when it is determined, according to the detected actual usage, that the hard disk corresponding to the initial capacity level exceeds a limit, searching for a hard disk corresponding to a second highest capacity level that is in the configured hard disk capacity levels and is higher than the initial capacity level; and when the hard disk corresponding to the second highest capacity level is obtained by means of searching, determining a migration ratio of a virtual node according to the initial capacity level, the second highest capacity level, and the actual hard disk capacity, and performing, according to the migration ratio, virtual node migration between hard disks that are corresponding to the configured hard disk capacity levels, where the virtual node carries stored data. By using the present invention, hard disk capacity heterogeneousness in massive storage can be supported.