Distributed RAID Virtual Chunk Mapping for Drive Enclosure Failure Tolerance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems lack DE-failure tolerance in distributed RAID schemes, which hinders high availability and rebuild speed.
Innovation Solution
A storage system that configures virtual chunks with user data and redundant data across multiple physical storage drives, ensuring that each virtual chunk is mapped to different physical drives within drive enclosures to maintain redundancy and availability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If drives are arranged in a distributed manner across multiple drive enclosures to achieve DE-failure tolerance, then high availability is improved, but rebuild speed deteriorates due to dispersed data locations
Solution Approach 1:
The patent segments virtual chunks into multiple virtual parcels and distributes them across different physical storage drives in different drive enclosures. This segmentation allows the system to maintain DE-failure tolerance by spreading data across enclosures while enabling parallel reconstruction of failed drives from multiple sources, thus resolving the contradiction between reliability and rebuild speed.
Solution Approach 2:
The patent introduces a new dimensional organization by mapping virtual parcels to physical drives across multiple drive enclosures in a structured manner. This multi-dimensional mapping approach allows data to be distributed for reliability while maintaining accessible reconstruction paths, overcoming the single-dimensional limitation of traditional RAID that either prioritized speed or reliability but not both simultaneously.
2Speed
If multiple virtual parcels from the same virtual chunk are stored on different physical storage drives, then rebuild speed is improved through parallel access, but device complexity increases due to mapping management
Solution Approach 1:
The patent introduces a controller as an intermediary that manages the mapping between virtual parcels and physical storage drives. This intermediary abstracts the complexity of distributed mapping management from the user and system operations, allowing parallel rebuild operations to occur while the controller handles the sophisticated mapping logic, thus resolving the contradiction between speed improvement and complexity increase.
Solution Approach 2:
The virtual chunk and virtual parcel structure serves multiple functions: it enables parallel rebuild operations, maintains DE-failure tolerance, and provides a standardized interface for data access. This multi-functionality justifies the mapping management complexity by delivering multiple performance benefits simultaneously, resolving the contradiction between speed and complexity.
3Reliability
If virtual parcels are distributed across k different physical storage drives, then data integrity is improved through redundancy, but storage space efficiency deteriorates due to replicated redundant data
Solution Approach 1:
The patent changes the organizational parameters of data storage by using virtual chunks and virtual parcels as intermediate units between logical volumes and physical drives. This parameter change allows flexible configuration of redundancy levels (k different drives) while optimizing the distribution of redundant data, achieving data integrity through redundancy while improving storage space efficiency compared to traditional RAID approaches.
Data Source
AI summary
It is made possible to pursue both higher speeds of rebuilds in the distributed RAID scheme and high availability due to acquisition of DE-failure tolerance. A virtual chunk includes k (k is an integer that is equal to or larger than two) virtual parcels including a virtual parcel having user data and a virtual parcel having element data that is redundant data for repairing the user data, the virtual parcels included in the same virtual chunk is stored by mapping the virtual parcels in storage areas of k mutually different physical storage drives among N (k<N) of the physical storage drives, and the maximum value of the numbers of the same virtual parcels to be mapped to the physical storage drives housed in the same drive enclosures is equal to or smaller than a predetermined value.