Data Pattern Identification for RAID Rebuild Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional RAID rebuild processes consume excessive physical storage media and create storage imbalances, leading to inefficient garbage collection in data storage devices using indirection systems, particularly in Solid-State Hybrid Drives (SSHDs) with Shingled Magnetic Recording (SMR), where data is not rewritten in the same physical location.
Innovation Solution
Implementing a data pattern identification process that recognizes and maps specific data patterns in a translation table, allowing unused portions to be distinguished from used ones, thereby avoiding unnecessary data generation during rebuilds and optimizing storage usage, and enhancing garbage collection efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional RAID rebuild processes are used in data storage devices with indirection systems, then data can be recovered, but excessive physical storage media is consumed and storage imbalances are created
Solution Approach 1:
The system performs preliminary identification of data patterns and their locations before the rebuild process begins. By pre-mapping data patterns to physical locations using the translation table, the system knows exactly which data to write and where, avoiding the conventional approach of rewriting all data and thereby conserving physical storage media.
Solution Approach 2:
The invention extracts and identifies specific data patterns from the storage media using pattern recognition techniques. By extracting only the necessary data patterns that need to be rebuilt rather than all data, the system reduces physical storage consumption during the rebuild process.
2Productivity
If data is not rewritten in the same physical location in SMR DSDs, then sequential writing can be maintained, but physical storage space is not conserved during maintenance operations
Solution Approach 1:
The system creates a mapping copy of data patterns to their physical locations in the translation table. This copy allows the system to identify and conserve physical storage space without disrupting the sequential writing capability of SMR DSDs, as the mapping enables precise tracking of data locations.
Solution Approach 2:
The translation table provides feedback about the physical locations of data patterns, enabling the system to make informed decisions about where to write data during maintenance operations. This feedback mechanism allows the system to conserve physical storage space while maintaining sequential writing.
3Quantity of substance
If garbage collection is performed to reclaim obsolete data, then storage space is recovered, but the process is delayed by storage imbalances
Solution Approach 1:
The system performs preliminary identification and mapping of data patterns before garbage collection. By knowing exactly which data is obsolete through pattern matching and translation table lookup, the system can immediately proceed with garbage collection without delays caused by storage imbalances or uncertainty about which data to reclaim.
4Ease of operation
If translation tables map all logical addresses to physical addresses, then data access is enabled, but storage efficiency decreases due to redundant mappings
Solution Approach 1:
Instead of maintaining full translation table mappings for all data, the system creates pattern copies - storing only the essential data pattern information and its location. This reduces the size of the translation table while maintaining data access capability, thereby improving storage efficiency.
Solution Approach 2:
The system performs preliminary pattern identification and stores only the necessary mapping information in the translation table. By pre-identifying data patterns and their locations, the system reduces redundant mappings while maintaining efficient data access through the condensed translation table.
Data Source
AI summary
Managing data stored in a Data Storage Device (DSD) including a memory. A translation table is maintained mapping logical addresses for data to physical addresses corresponding to a location where the data is stored in the memory. A data pattern is identified of particular data to be stored in the memory or of particular data already stored in the memory, and it is indicated in the translation table that at least one logical address for the particular data is associated with the data pattern.