Nonvolatile Memory RAID Parity for Hot Block Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional nonvolatile memory devices, such as flash memory, face data loss and reliability issues due to limited program/erase endurance and electrical malfunctions, which existing error correction techniques fail to adequately address, especially in hot memory blocks that experience frequent use and deterioration.
Innovation Solution
Implementing a RAID technique in nonvolatile memory devices to identify hot memory blocks and generate multiple units of RAID parity data using data from these blocks, ensuring data reliability and recovery even in cases of read failures across multiple units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional error correction techniques are used in nonvolatile memory devices, then device complexity is reduced, but data reliability deteriorates in hot memory blocks
Solution Approach 1:
The patent applies preliminary action by generating RAID parity data in advance for hot memory blocks before actual data storage operations occur. The controller identifies hot memory blocks that experience frequent program/erase cycles and pre-generates parity data using RAID algorithms, so that when read failures occur, recovery can immediately utilize the pre-computed parity data without requiring complex real-time error correction mechanisms.
Solution Approach 2:
The patent introduces RAID parity data as an intermediary element between the hot memory blocks and the data recovery process. Instead of directly implementing complex error correction in the memory device, the system uses RAID parity data stored in separate locations as a mediator to recover lost data, thereby reducing the complexity of error correction mechanisms within the memory device itself.
2Reliability
If RAID parity data is generated for hot memory blocks, then data recovery capability is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by implementing RAID parity generation selectively only for hot memory blocks that experience frequent program/erase cycles, rather than applying it uniformly to all memory blocks. The controller identifies specific hot blocks based on their access patterns and applies the RAID protection mechanism locally to these high-risk areas, thereby improving data recovery capability where needed while minimizing the overall complexity increase in the controller.
3Reliability
If multiple RAID parity data units are generated from hot pages, then resistance to read failures is improved, but processing time increases
Solution Approach 1:
The patent generates multiple units of RAID parity data from hot pages in advance, during initial programming operations or during idle periods, rather than generating them on-demand when read failures occur. This preliminary generation approach ensures that when read failures happen, the system can immediately use the pre-computed parity data for recovery, thereby improving read failure resistance without incurring time delays during critical recovery operations.
Data Source
AI summary
A method of operating a nonvolatile memory device comprising a plurality of memory blocks comprises storing first data and second data to be stored in a hot memory block of the memory blocks in a first buffer, transferring the first data stored in the first buffer to a second buffer to program the first data in the hot memory block, and generating RAID parity data based on the first and second data, wherein the RAID parity data and the first data form part of the same write stripe.