SSD Metadata Storage in SLC Regions for NAND Reliability

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing SSDs lack a mechanism to differentiate between metadata and normal data storage, leading to poor storage reliability due to the use of TLC Nand flash, which results in instability and potential file system crashes.

Innovation Solution

Implementing a system that segregates metadata storage to a Single-Level Cell (SLC) memory mode within the SSD, ensuring metadata is stored in a more reliable region, backed up if necessary, and utilizing optimization techniques like RAID to enhance data integrity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data is stored in SSD using traditional methods without addressing NAND wear, then storage operations can proceed quickly, but the SSD lifespan is reduced due to excessive wear on NAND blocks

Engineering Contradiction:
ImproveSSD lifespanVSAvoidwear leveling complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary actions by pre-mapping logical block addresses to physical block addresses before data is actually written. The wear leveling manager proactively manages block allocation and tracking in advance, distributing write operations across multiple blocks before any single block becomes full or worn, thereby extending SSD lifespan without requiring complex real-time wear leveling operations

Inventive Principle:
Principle #10Preliminary action

2Ease of operation

If the SSD controller manages wear leveling independently without host involvement, then the SSD can operate autonomously, but the host system cannot track which logical blocks have been moved to different physical locations

Engineering Contradiction:
Improvehost awareness of data locationVSAvoidmapping table management
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The patent merges the wear leveling management functionality into the host system's file system or storage management layer. The host maintains a mapping table that tracks logical block addresses to physical block addresses, allowing the host to be aware of data location changes while the SSD controller focuses on physical write operations. This division of labor simplifies the SSD controller's complexity while providing the host with the information it needs

Inventive Principle:
Principle #5Merging (Combining)

3Reliability

If the SSD uses a fixed mapping between logical and physical blocks, then address translation is simple, but NAND blocks cannot be reused after wear, reducing overall storage capacity and lifespan

Engineering Contradiction:
ImproveNAND block reuse capabilityVSAvoidaddress mapping complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system implements dynamic address mapping where the mapping between logical block addresses and physical block addresses can change over time. When a physical block becomes full or worn, the system dynamically remaps subsequent logical blocks to different physical blocks, allowing previously used physical blocks to be erased and reused. This dynamic adaptation extends SSD lifespan while maintaining efficient space utilization

Inventive Principle:
Principle #15Dynamics

4Reliability

If the SSD controller implements comprehensive wear leveling algorithms, then NAND block wear is distributed evenly, but the controller's processing burden and power consumption increase

Engineering Contradiction:
ImproveNAND wear distributionVSAvoidcontroller power consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system applies partial wear leveling by implementing wear leveling only at the granularity of logical block addresses rather than individual physical blocks. The wear leveling manager distributes logical block allocations across physical blocks to prevent any single block from becoming excessively full or worn, but avoids the computational overhead of managing individual block-level wear leveling. This approach achieves sufficient wear distribution while minimizing controller processing burden and power consumption

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3765953B1Systems and methods for storing data in SSD
Publication Date: 2026.05.20 ZHEJIANG DAHUA TECH CO LTD
  • EP3765953B1 patent drawingFigure 1
  • EP3765953B1 patent drawingFigure 2~4
  • EP3765953B1 patent drawingFigure 5

AI summary

System and method for storage data in SSD may be provided. The method may include receiving data writing feature information sent by a file system during an initialization process. The method may include determining, based on the data writing feature information, a size of metadata storage space corresponding to the metadata. The method may further include determining, based on the size of metadata storage space, a target storage region for storing the metadata in the SSD.