Hybrid MRAM-NAND Storage Controller for Mobile Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mass storage devices rely on NAND flash memories, which are slow, unreliable, and have limited endurance, making them unsuitable for critical data management due to their block-based organization, limited erase cycles, and requirement for out-of-place updates, leading to performance overhead and unpredictable latency.
Innovation Solution
A mass storage device is designed with a hybrid architecture that combines NAND flash and MRAM, where MRAM is used for critical data requiring high performance and reliability, and NAND flash for non-critical data, allowing for in-place updates and reducing the need for garbage collection, thereby enhancing system performance and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If NAND flash memory is used for storage, then large storage capacity at reasonable price is achieved, but performance and reliability are insufficient due to slow access speed and limited endurance
Solution Approach 1:
The storage device is segmented into two distinct memory types: NAND flash memory for storing non-critical user data requiring large capacity, and MRAM for storing critical data requiring high reliability and performance. This segmentation allows each memory type to be optimized for its specific function, resolving the contradiction between capacity and reliability.
Solution Approach 2:
Different quality levels of memory are applied to different data types: high-reliability MRAM is used locally for critical data (controller private data, security keys, frequently accessed host data), while standard NAND flash is used for non-critical data. This local quality differentiation ensures that reliability is enhanced where needed without sacrificing overall storage capacity.
2Ease of manufacture
If NAND flash memory is used, then cost-effectiveness is maintained, but performance overhead and unpredictable latency occur due to block-based organization and garbage collection requirements
Solution Approach 1:
The storage system is divided into two functional segments: NAND flash for bulk storage operations and MRAM for performance-critical operations. By segmenting the workload, the system achieves both cost-effectiveness (using inexpensive NAND for most data) and high performance (using fast MRAM for critical data), eliminating the performance overhead inherent in pure NAND implementations.
Solution Approach 2:
MRAM acts as an intermediary memory layer between the controller and NAND flash. It handles frequently accessed data and critical operations, mediating between the host and the slower NAND flash storage, thereby improving overall system performance and latency predictability while maintaining cost-effectiveness.
3Adaptability or versatility
If frequent erase operations are performed on flash memory, then data updates are possible, but the lifespan of the flash memory is reduced due to limited erase cycles
Solution Approach 1:
The update-intensive workload is extracted from NAND flash and relocated to MRAM. MRAM handles all frequent writes and updates due to its unlimited write endurance, while NAND flash is used only for sequential programming operations. This extraction preserves the lifespan of NAND flash memory while maintaining full data update capability through the MRAM component.
Solution Approach 2:
Different durability characteristics are applied to different memory regions: MRAM with unlimited write endurance is used for frequently updated critical data, while NAND flash with limited erase cycles is used for static or rarely modified data. This local quality differentiation allows frequent updates without reducing the overall lifespan of the storage system.
4Ease of operation
If out-of-place updates are used in flash memory, then data can be written to erased areas, but garbage collection overhead and unpredictable operational latency increase
Solution Approach 1:
The garbage collection function is extracted from the system by using MRAM for all write operations. Since MRAM supports in-place updates without requiring erasure, the complex garbage collection process inherent to flash memory is eliminated entirely, removing the associated performance overhead and unpredictable latency.
Solution Approach 2:
Different update mechanisms are applied to different data types: MRAM provides in-place update capability for critical data requiring frequent modifications, while NAND flash uses traditional out-of-place updates for non-critical data. This local quality differentiation ensures that performance-critical operations benefit from fast in-place updates without compromising the overall write flexibility of the system.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The hybrid approach improves system performance by eliminating the need for garbage collection and wear leveling, increases reliability, and extends the lifespan of storage media, while maintaining cost-effectiveness by utilizing NAND flash for non-critical data.
Implementation Method 1
management of the memory array of storage device by a controller using magnetic random access memory (MRAM)
Data Source
AI summary
A mass storage device includes a controller configured to communicate with a host. The controller is coupled to a first memory and a second memory, the first and second memories being of different types. The mass storage device includes a storage media partitioned into a plurality of Logical Units (LUNs) based on capabilities and resources of the mass storage device. The mass storage device further includes the first memory and the second memories and a hybrid reserved area spanning at least a portion of the first and second memories.