Flash Translation Layer Log and Partial Block Data Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-capacity flash memory devices face reliability issues due to integration limitations, making it difficult to determine and address malfunctioning memory blocks, leading to potential data loss and performance degradation.
Innovation Solution
A data management method for a flash translation layer that determines data size and stores it in either a log block or a partial block, with separate merge operations and mapping tables for efficient data management, reducing the need for frequent merge operations and avoiding write timeouts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is stored in traditional memory blocks with fixed write units, then storage capacity is maximized, but write operations timeout when data size is smaller than write unit
Solution Approach 1:
The patent segments the storage structure into two distinct types: log blocks for full-page writes and partial blocks for partial-page writes. This segmentation allows each block type to be optimized for its specific use case, preventing write timeouts by matching the data size to the appropriate block type rather than forcing all writes into fixed-size blocks.
Solution Approach 2:
The patent introduces dynamic block management where the system can switch between log blocks and partial blocks based on the size of incoming data. This dynamic approach allows the storage system to adapt to varying data sizes, performing incremental merges for partial blocks when thresholds are met, thereby avoiding write timeouts while maintaining data integrity.
2Stability of the object's composition
If merge operations are performed frequently to maintain data organization, then data organization is improved, but write performance degrades
Solution Approach 1:
The patent implements periodic merge operations for partial blocks based on a threshold mechanism. Instead of performing merges after every write operation, the system accumulates partial writes in partial blocks and performs incremental merges only when the data volume reaches a predetermined threshold. This periodic approach reduces the frequency of merge operations, thereby improving write performance while still maintaining data organization.
Solution Approach 2:
The patent applies partial merge operations only when necessary - specifically when partial blocks reach their data threshold. This partial action approach avoids unnecessary full merge operations, reducing the overhead on write performance while still achieving the necessary data organization and consolidation at appropriate intervals.
3Reliability
If replacement blocks are reserved for malfunctioning memory blocks, then reliability is improved, but storage capacity is reduced
Solution Approach 1:
The patent implements a self-service replacement mechanism where partial blocks that have accumulated sufficient data through incremental merges automatically serve as replacement blocks when malfunctioning blocks are detected. This self-service approach eliminates the need for pre-reserved replacement blocks, thereby maintaining reliability while maximizing usable storage capacity as all blocks can be utilized for data storage.
Data Source
AI summary
A data management method includes determining the size of input data, storing the input data in a log block if the size of the input data is determined to be a write unit, and storing the input data in a partial block if the size of the input data is determined to be smaller than the write unit. The log block is a temporary block storing data of same addresses and the partial block is a temporary block storing data regardless of their addresses. The memory system includes a nonvolatile memory and a memory controller configured to control the nonvolatile memory. The memory controller is configured to temporarily store input data smaller than a write unit in a selected memory block even when the input data have different addresses.