Offloading RAID Processing to Storage Enclosure Controllers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems face inefficiencies in handling duplicate data blocks and performance bottlenecks due to address-based mapping, which can lead to increased storage capacity usage and reduced performance during duplicate block writes and RAID processing.
Innovation Solution
A storage system with a content-addressable architecture that offloads RAID processing, compression, and error handling to processing devices within storage enclosures, allowing for efficient data management and reduced load on the storage controller, enabling improved storage capacity and performance by calculating RAID parity and rebuilding data on failed drives within the enclosures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If address-based mapping is used for data storage and retrieval, then the storage system can maintain simple address-based functionality, but storage capacity efficiency decreases and performance degrades due to duplicate block writes
Solution Approach 1:
The storage system performs self-service by automatically identifying duplicate data blocks through content-based hashing and eliminating redundant storage. The system autonomously computes hash values for incoming data blocks, checks for duplicates in the namespace, and prevents duplicate writes without requiring external intervention or manual management.
Solution Approach 2:
The patent changes the fundamental parameter of data identification from address-based to content-based hashing. By computing hash values from data content and using these hashes as the basis for storage decisions, the system transforms how data is mapped and managed, enabling automatic duplicate detection and elimination while improving storage efficiency.
2Productivity
If the storage controller handles all RAID processing and error handling, then centralized control is maintained, but the storage controller becomes a bottleneck reducing system performance
Solution Approach 1:
The patent segments RAID processing and error handling functions from the centralized storage controller and distributes them to individual storage enclosure processing devices. Each enclosure's processing device independently handles RAID operations for its local drives, including parity calculation, data reconstruction, and error correction, eliminating the controller bottleneck.
Solution Approach 2:
The processing devices in storage enclosures act as intermediaries between the storage controller and the RAID drives. They receive commands from the controller and autonomously execute complex RAID operations, buffering and mediating the interaction to prevent controller overload while maintaining coordinated system operation.
3Quantity of substance
If duplicate data blocks are stored separately, then address-based mapping is simple to implement, but storage capacity is wasted and write performance decreases
Solution Approach 1:
The system uses content-based hashing to create a digital fingerprint (hash value) of each data block and stores this hash in a namespace for quick comparison. When a new block arrives, its hash is computed and compared against existing hashes to identify duplicates, allowing the system to recognize and eliminate redundant copies without complex management operations.
Solution Approach 2:
The patent fundamentally changes the identification parameter from physical address to content-based hash value. This transformation enables automatic duplicate detection by comparing hash values rather than managing address mappings, improving both storage capacity utilization and operational simplicity through automated content-based deduplication.
Data Source
AI summary
A storage system comprises a plurality of enclosures and a storage controller. Each enclosure comprises at least one processing device and a plurality of drives configured in accordance with a redundant array of independent disks (RAID) arrangement. The storage controller obtains an input-output request to read a target data page and issues a command to the at least one processing device of a given enclosure. Responsive to receiving the command, the at least one processing device of the given enclosure obtains data pages associated with a given stripe from the plurality of drives of the given enclosure. The at least one processing device rebuilds the given stripe based at least in part on the obtained data pages, obtains the target data page from the rebuilt given stripe and provides the obtained target data page to the storage controller.