Relocating Highly Read Data to Low Impact Pages
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional memory sub-systems experience increased bit error rates and reduced reliability due to read disturb effects, particularly when highly read data is stored in pages that suffer more read stress, leading to a shorter lifespan and decreased performance.
Innovation Solution
A highly read data manager identifies and relocates highly read data to low impact read disturb pages, reducing the voltage differences between adjacent wordlines and mitigating read disturb effects by intelligently selecting target wordlines for data storage during relocation and write operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If highly read data is stored in pages that suffer more read stress, then data accessibility is improved, but bit error rate increases and reliability decreases
Solution Approach 1:
The patent applies local quality by differentiating between high read disturb pages and low read disturb pages within the memory device. Highly read data is specifically placed in low read disturb pages, which have different characteristics (lower vulnerability to read disturb effects) compared to regular pages. This localized optimization of storage locations resolves the contradiction by maintaining fast accessibility while protecting against bit errors.
Solution Approach 2:
The patent implements preliminary action by proactively identifying highly read data and relocating it to low read disturb pages before read disturb effects can cause significant bit errors. The controller monitors read patterns and performs preventive relocation, rather than waiting for errors to occur. This advance action maintains reliability while preserving data accessibility.
2Speed
If highly read data is stored in pages that suffer more read stress, then read operations can proceed quickly, but read disturb impact increases and lifespan decreases
Solution Approach 1:
The patent creates distinct storage zones within the memory device: low read disturb pages optimized for highly read data and regular pages for other data. This local differentiation allows read operations to proceed quickly on highly accessed data while protecting the overall memory device lifespan by concentrating read stress on pages designed to handle it.
Solution Approach 2:
The patent converts the harmful effect of read disturb into a beneficial strategy by identifying and utilizing low read disturb pages. Instead of viewing read disturb as purely detrimental, the system leverages the natural variation in read disturb characteristics across different pages to extend memory lifespan while maintaining fast read performance for frequently accessed data.
3Reliability
If data relocation is performed frequently to manage read disturb, then reliability is improved, but write amplification increases
Solution Approach 1:
The patent reduces write amplification by performing preliminary relocation of highly read data to low read disturb pages during initial write operations or during off-peak times. By proactively placing data in optimal locations beforehand, the system minimizes the need for frequent corrective relocations, thus maintaining data integrity while reducing unnecessary write operations.
Solution Approach 2:
The patent implements feedback mechanisms that monitor read patterns and read disturb levels to intelligently determine when relocation is necessary. This feedback-driven approach prevents excessive relocation operations by only moving data when actually needed, balancing reliability improvement with write amplification control.
Data Source
AI summary
A highly read data manager of a memory device receives a request to perform receives a request to perform a data relocation operation on a first wordline of a plurality of wordlines for a memory device, the memory device comprising a plurality of multi-level memory cells, wherein each multi-level memory cell comprises a plurality of pages; determines at the first wordline comprises data stored at one or more high read disturb pages of the plurality of pages; determines whether the data comprises a characteristic that satisfies a threshold criterion in relation to additional data stored on additional wordlines of the plurality of wordlines; responsive to determining that the data comprises the characteristic that satisfies the threshold criterion, identifies one or more low read disturb pages of the plurality of pages of a target wordline for relocating the data; and responsive to identifying the one or more low read disturb pages of the target wordline, stores at least a portion of the data at the one or more low read disturb pages of the target wordline.