NAND Flash Block Segmentation for SSD I/O Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current storage systems, particularly SSDs, face challenges in enhancing data input/output performance due to limitations in efficiently utilizing physical resources, leading to suboptimal performance in managing data access and storage operations.
Innovation Solution
The implementation of a storage system that classifies blocks into first and second blocks, with the controller performing read, write, or erase operations based on physical or virtual block addresses, utilizing a flash translation layer for efficient data management, wear leveling, and garbage collection to optimize NAND memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the storage system uses traditional block management methods, then the structure is simple, but the data input/output performance is insufficient
Solution Approach 1:
The storage system divides blocks into multiple types (first blocks and second blocks) with different management methods. First blocks are managed at the block level while second blocks are managed at the page level within block groups. This segmentation allows the system to optimize I/O performance for different data access patterns without complicating the entire block management structure.
Solution Approach 2:
The patent introduces a new dimension of management granularity by organizing second blocks into block groups that can be managed collectively at the page level. This adds a intermediate management layer between individual blocks and the entire storage device, enabling more flexible I/O operations and improving overall data input/output performance.
2Speed
If the storage system implements fine-grained block management, then the data access performance is improved, but the wear leveling efficiency decreases
Solution Approach 1:
The storage system applies different management strategies to different types of blocks. First blocks use traditional block-level management suitable for sequential access, while second blocks use page-level management within block groups optimized for random access. This local quality differentiation allows the system to achieve high data access speed for appropriate workloads while maintaining overall wear leveling efficiency through selective application of management techniques.
Data Source
AI summary
According to one embodiment, a storage system includes a nonvolatile memory and a controller. The controller classifies blocks in the nonvolatile memory into a plurality of first blocks and a plurality of second blocks organized as a plurality of block groups each including a set of blocks. The controller receive a first read, write or erase command from a host, and performs a read, write or erase operation on one first block in the first blocks that is designated by a physical address of the first read, write or erase command. The controller receive a second read, write or erase command from the host, and performs a read, write or erase operation on blocks in one block group in the block groups that is designated by a physical address of the second read, write or erase command.