Two-Layer Memory Coding With Dynamic Layout to Avoid Write Collisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory sub-systems face media access collisions and increased buffer lifetimes due to fixed media layouts, which can lead to inefficiencies in data storage and retrieval, especially when multiple write streams are active.
Innovation Solution
Implementing dynamic data placement and a two-layer code with low parity cost, where the media layout is determined based on the availability of integrated circuit dies at the time of input/output scheduling, and using a combination of error correcting and erasure codes to encode data for storage across multiple planes, avoiding collisions and optimizing storage capacity utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed media layout is used, then the media layout is simple to implement, but media access collisions occur and buffer lifetime increases
Solution Approach 1:
The patent implements dynamic data placement where the media layout is determined based on the availability of integrated circuit dies at the time of input/output scheduling, rather than using a fixed predetermined layout. This dynamic approach adapts to changing system conditions to optimize performance while avoiding media access collisions.
2Ease of operation
If a fixed media layout is used, then the system is easier to manage, but buffer lifetime increases and performance decreases
Solution Approach 1:
The system dynamically determines media layout at I/O scheduling time based on die availability, which reduces buffer lifetime by enabling more efficient data placement and reducing the time data needs to wait in buffers before being written to memory.
3Reliability
If storage capacity is over-provisioned to handle wear leveling, then reliability improves, but storage space amplification increases
Solution Approach 1:
The patent applies different coding schemes (error correcting codes and erasure codes) to different portions of data based on their specific requirements. Error correcting codes are used for data requiring high reliability, while erasure codes are used for other data, optimizing the balance between reliability and storage efficiency locally rather than applying a uniform approach.
Solution Approach 2:
The system changes coding parameters dynamically by selecting between different code types and configurations based on data characteristics and system conditions, allowing optimization of both reliability and storage space utilization without over-provisioning.
Data Source
AI summary
A memory sub-system configured to encode data using an error correcting code and an erasure code for storing data into memory cells and to decode data retrieved from the memory cells. For example, the data units of a predetermined size are separately encoded using the error correcting code (e.g., a low-density parity-check (LDPC) code) to generate parity data of a first layer. Symbols within the data units are cross encoded using the erasure code. Parity symbols of a second layer are calculated according to the erasure code. A collection of parity symbols having a total size equal to the predetermined size can be further encoded using the error correcting code to generate parity data for the parity symbols.