SSD Cache Partition Mirroring for Faster Small-Block Writes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The low response efficiency in writing data to solid state disks due to the need to write random small data blocks one by one, leading to inefficient response times and user experience degradation in hyper-converged systems.
Innovation Solution
A data writing method that preprocesses data blocks through segmentation and integration, mirrors them to a cache partition, and then writes them to a storage partition, using mirroring and fault tolerance to ensure rapid and reliable data transfer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data blocks are written one by one according to writing requirements, then data writing reliability is ensured, but response efficiency deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-processing data blocks before actual writing operations. Data blocks are segmented or integrated in advance to form optimal writing units, and mirroring operations are performed beforehand to prepare redundant copies. This preprocessing reduces the complexity and time of actual writing operations while ensuring reliability through pre-established mirror copies.
Solution Approach 2:
The patent implements copying through a mirroring mechanism where data blocks are written to multiple partitions simultaneously. When data is written to one partition, mirror copies are created in other partitions, enabling fault tolerance and parallel writing operations. This copying approach allows the system to maintain reliability through redundancy while improving response efficiency through parallel processing.
2Reliability
If random small data blocks are written sequentially, then data integrity is maintained, but write amplification increases
Solution Approach 1:
The patent applies merging by integrating multiple small data blocks into larger writing units through the integration operation. Small data blocks that would otherwise require separate writing operations are combined into unified writing sequences, reducing the total number of writing operations and eliminating write amplification while maintaining data integrity through coordinated mirroring of the integrated blocks.
Solution Approach 2:
The patent uses preliminary action by performing segmentation and integration operations before writing. Data blocks are pre-processed to determine optimal writing sequences and groupings, allowing the system to write larger consolidated units rather than individual small blocks. This preprocessing reduces write amplification by minimizing the number of separate writing operations required while ensuring data integrity through pre-planned mirroring strategies.
3Reliability
If mirroring operations are performed for all data blocks, then fault tolerance is improved, but system complexity increases
Solution Approach 1:
The patent applies local quality by implementing mirroring selectively rather than uniformly across all data blocks. The system determines which data blocks require mirroring based on their characteristics, importance, or access patterns, and applies mirroring operations only to those specific blocks. This selective approach maintains fault tolerance for critical data while reducing overall system complexity by avoiding unnecessary mirroring operations for less critical data.
Solution Approach 2:
The patent implements partial action by performing mirroring operations on only a subset of data blocks rather than all blocks. The system identifies critical data blocks that benefit most from fault tolerance and applies mirroring selectively to those blocks, achieving adequate fault tolerance with reduced complexity compared to universal mirroring. This partial approach balances reliability requirements with system simplicity.
Data Source
AI summary
A data writing method and apparatus, a solid state drive, an electronic device, and a non-transitory readable storage medium are provided. According to the method, data to be written is preprocessed. The preprocessed data blocks are mirrored to a first partition of a target solid state drive. Furthermore, response information of a mirroring result is sent to the target object, and then the preprocessed data blocks are written to a second partition of the target solid state drive from the first partition.