US Patent Application 17828077. BALANCED DATA MIRRORING DISTRIBUTION FOR PARALLEL ACCESS simplified abstract
Contents
BALANCED DATA MIRRORING DISTRIBUTION FOR PARALLEL ACCESS
Organization Name
Inventor(s)
BALANCED DATA MIRRORING DISTRIBUTION FOR PARALLEL ACCESS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17828077 titled 'BALANCED DATA MIRRORING DISTRIBUTION FOR PARALLEL ACCESS
Simplified Explanation
The abstract of the patent application describes a method for distributing mirrored pairs in a RAID-1 (Redundant Array of Independent Disks) system in a balanced and deterministic way to increase data access parallelism.
- The method is applicable to a group of k+1 disks organized in a matrix structure, with each disk divided into k same-size subdivisions.
- Mirrors of data members on the first disk are distributed across all other disks along a matrix diagonal, where the row index matches the column index for each mirror.
- Additional mirror pairs are created and distributed symmetrically in two submatrix triangles, separated by the diagonal.
- The mirrors in the triangles are flipped around the matrix diagonal, allowing for easy swapping of row and column indices to find corresponding mirror data in the other triangle.
- This distribution method improves data access parallelism in a RAID-1 system, enhancing performance and redundancy.
Original Abstract Submitted
Mirrored pairs in a RAID-1 are distributed in a balanced and deterministic way that increases data access parallelism. For a group of k+1 disks that can be represented as a matrix of disk rows indexed 0 through k, where each disk is organized into k same-size subdivisions, in columns indexed 1 through k, corresponding mirrors of data members on the first disk (row index 0) are distributed across all other disks along a matrix diagonal such that the row index is the same as the column index for each mirror. Additional mirror pairs are created and symmetrically distributed in two submatrix triangles that are defined and separated by the diagonal. The two triangles are populated with symmetrically distributed mirrors that are flipped around the matrix diagonal such that for any mirror data in one triangle, its corresponding mirror data can be found in the other triangle by swapping the row and column indices.
