Zoned UFS Address Mapping for Lower Write Amplification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The implementation of zoned storage in memory systems, particularly in zoned universal flash storage (zoned UFS), is challenging due to the need for efficient management of data access patterns and the lack of compatibility with existing memory system controllers.
Innovation Solution
The method involves dividing the memory system into zones that support sequential read and write operations, using a mapping relationship to determine physical page addresses based on logical addresses, and employing a two-level mapping table to facilitate efficient data operations, including garbage collection and data management across different zones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the memory system divides storage space into zones and uses sequential read/write operations, then write amplification is reduced and memory life is extended, but the device complexity increases due to the need for zone management and mapping relationships
Solution Approach 1:
The storage space is divided into multiple zones with different access patterns (sequential zones and random zones). Each zone is managed independently with its own mapping relationship, allowing the system to optimize for sequential operations in zoned areas while maintaining random access capability in other areas, thus extending memory life through reduced write amplification without requiring complete system redesign
Solution Approach 2:
A mapping relationship table is introduced as an intermediary layer between logical addresses and physical storage locations. This mapping table manages the zone divisions and sequential/random access patterns, abstracting the complexity of zone management from the core memory operations and enabling flexible zone management without increasing fundamental device complexity
2Productivity
If a mapping relationship is used to translate logical addresses to physical page addresses in zoned storage, then data access patterns are optimized, but the device complexity increases due to additional mapping tables and translation operations
Solution Approach 1:
The mapping relationship is segmented into zone-level mappings and page-level mappings. The zone mapping table stores the relationship between logical addresses and zone identifiers, while the page mapping table stores the relationship between zones and physical page addresses. This segmentation allows efficient data access by breaking down the translation process into manageable steps without requiring a single complex mapping table
Solution Approach 2:
The mapping system transitions from a traditional two-level mapping (logical address to physical address) to a three-level mapping structure (logical address to zone identifier, zone identifier to physical page address). This dimensional change enables optimized data access patterns by introducing zone as an intermediate dimension, allowing sequential operations within zones while maintaining overall system efficiency
3Loss of energy
If the storage space is divided into zones with sequential read/write support only, then write amplification is reduced, but the adaptability decreases because random read/write operations are not supported in zoned areas
Solution Approach 1:
The storage space is segmented into different types of zones: sequential zones that support only sequential read/write operations to minimize write amplification, and random zones that support random read/write operations for flexibility. This segmentation allows the system to reduce write amplification in appropriate areas while maintaining adaptability in other areas, resolving the contradiction between energy efficiency and versatility
Solution Approach 2:
Different access patterns are applied locally to different zones based on their intended use. Sequential zones are optimized for sequential operations with restricted access patterns to reduce write amplification, while random zones maintain full random read/write capability. This local differentiation allows the system to achieve low write amplification where needed without sacrificing overall adaptability
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Examples of the present application relate to a method of operating a memory system, a controller, a memory system and an electronic device, and relate to but are not limited to the field of memory technology. The method includes: in response to a first command comprising a first logical address of first data on which an operation is to be performed is received, determining at least one zone corresponding to the first logical address, the at least one zone corresponding to a first storage space of a memory of the memory system; obtaining at least one first physical page address corresponding to the at least one zone according to a first mapping relationship and the at least one zone determined; and performing the operation on the first data in the first storage space according to the at least one first physical page address, wherein a storage capacity of one of the zones is less than a storage capacity corresponding to a physical block of the memory, and the first storage space is a storage space that only supports sequential read and sequential write. In this way, zoned storage of the memory system can be implemented.