Granular RAID Scaling via Modulo Partition Relocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems face inefficiencies in scaling storage capacity without introducing excess unused capacity, particularly as individual drive storage costs increase with technological advancements.
Innovation Solution
The method involves creating indexed partitions on (D+P) drives, forming protection groups with D data members and P parity members, and using modulo arithmetic to select and relocate protection group members to new drives, allowing for incremental and granular scaling and splitting of drive arrays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If storage capacity is increased by adding new protection groups with (D+P) drives, then total storage capacity is improved, but excess unused capacity and cost increase
Solution Approach 1:
The patent segments the protection group members across multiple drives using indexed partitions. Each drive is divided into M*(D+P) partitions, allowing individual partition-level management. This segmentation enables granular allocation where new drives can be added and specific partitions migrated, rather than requiring complete (D+P) drive sets, thus eliminating excess capacity waste.
Solution Approach 2:
The patent introduces a partition index dimension alongside the traditional drive structure. By organizing data across drives[D][P] with M partitions each, the system adds a partition index dimension that enables fine-grained data redistribution. This dimensional expansion allows incremental capacity addition by migrating individual partitions to new drives rather than adding complete protection groups.
2Quantity of substance
If storage capacity is increased by adding (D+P) new drives, then total storage capacity is improved, but the scaling increment is too large and inflexible
Solution Approach 1:
The protection group is segmented into individual partitions across multiple drives. Instead of treating the entire (D+P) drive set as a single unit, the system divides it into M partitions per drive, enabling incremental migration of individual partitions to new drives. This allows flexible scaling where capacity can be increased by adding just one or a few drives and migrating specific partitions, rather than requiring complete (D+P) drive increments.
Solution Approach 2:
The system implements dynamic partition migration using modulo arithmetic to determine which protection group members relocate to new drives. As new drives are added, the modulo operation dynamically calculates redistribution patterns, enabling the system to adapt its structure flexibly rather than following fixed (D+P) increment rules. This dynamic approach allows continuous adjustment of capacity scaling based on actual needs.
3Productivity
If protection group members are redistributed to new drives, then capacity utilization is improved, but data relocation complexity increases
Solution Approach 1:
The patent changes the organizational parameters from traditional drive-based protection groups to partition-indexed distribution across drives. By introducing partition indices and using modulo arithmetic operations, the system transforms the relocation process into a parameter-driven calculation rather than manual restructuring. This parameter change simplifies the complexity by providing automated calculation rules for data migration.
Solution Approach 2:
The system implements self-service data relocation through automated modulo arithmetic calculations. When new drives are added, the system automatically determines which protection group members need to relocate and where they should be placed, without requiring manual intervention for complex redistribution decisions. The modulo operation self-manages the allocation logic, reducing operational complexity despite the increased capacity utilization requirements.
Data Source
AI summary
A subset of drives with protection groups that have D data members and P parity members is scaled in single drive increments. A plurality of indexed partitions equal in size and number are created on (D+P) indexed drives. Protection groups that have D data members and P parity members are created on individual partition indexes of the (D+P) drives. When a new drive is added some of the protection group members located on the (D+P) drives are selected using modulo arithmetic. The selected protection group members are relocated to the new drive and new protection groups are created using the partitions made available due to the relocations. When (D+P) new drives have been added the drive subset may be split into two drive subsets. The modulo arithmetic may include selecting the members of protection group X that are on partition X of drive Y that satisfy the condition (X−Y+N−1) modulo W=0, where N is a count of new drives added to the (D+P) drives and W=(D+P). Alternatively, the modulo arithmetic may include selecting the members of protection group X on partition X of drive Y that satisfy the condition (X+Y−N−1) modulo W=0.