SSD Erasable Unit Remapping for Dynamic Over-Provisioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solid-state drives (SSDs) face inefficiencies in managing data storage operations due to the need for large over-provisioning spaces and sequential program/erase (P/E) operations, which hinder performance.
Innovation Solution
Implement minimal initial over-provisioning and dynamically manage erasable units of memory blocks by remapping and moving unused blocks offline to reserve space on demand, allowing concurrent P/E operations on adjacent units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a sufficiently large over-provisioning space is reserved to guarantee sufficient blocks for worst-case customer workloads, then reliability is improved, but device capacity is reduced
Solution Approach 1:
The patent implements dynamic over-provisioning management where the system monitors workload patterns and dynamically adjusts the over-provisioning space allocation. Instead of reserving a fixed large portion for worst-case scenarios, the system adaptively manages which blocks are available for host use versus reserved for over-provisioning, allowing optimal capacity utilization while maintaining reliability guarantees when needed
Solution Approach 2:
The system changes the parameter of over-provisioning space allocation from a static fixed value to a dynamic variable that adjusts based on actual workload conditions. By monitoring usage patterns and block availability, the system modifies the effective over-provisioning ratio in real-time, reducing it during normal operation to maximize capacity and increasing it only when reliability requirements demand
2Ease of manufacture
If program/erase operations are implemented successively on corresponding blocks in a superblock, then manufacturing constraints are satisfied, but productivity is reduced
Solution Approach 1:
The patent segments the superblock structure into independently manageable erasable units that can be operated on separately. Instead of treating the entire superblock as a single unit requiring sequential operations, the system divides it into smaller erasable units that can be processed in parallel, allowing simultaneous P/E operations on different units while still respecting the superblock organizational constraints
Solution Approach 2:
The system introduces a new dimension of parallelism by enabling concurrent P/E operations across multiple erasable units within the superblock structure. While maintaining the hierarchical superblock organization, the patent allows operations to proceed simultaneously on different units, effectively adding a temporal parallelism dimension that was previously constrained by sequential processing requirements
3Reliability
If adjacent blocks within the same superblock cannot be reclaimed, then reliability is maintained, but loss of time increases due to inability to implement concurrent operations
Solution Approach 1:
The patent segments the superblock into independent erasable units that can be individually reclaimed and operated on. This segmentation allows the system to maintain reliability guarantees for each unit while enabling concurrent reclamation operations on multiple units simultaneously, eliminating the time loss associated with sequential processing of adjacent blocks
Data Source
AI summary
This invention is directed to managing data storage in a memory system. The memory system receives a host request for an erase operation on a first erasable unit of a first memory block. In response to the host request, the memory system identifies an unused erasable unit of a second memory block in the memory system, remaps the first erasable unit of the first memory block to the unused erasable unit of the second memory block, and applies the erase operation on the unused erasable unit of the second memory block in place of the first erasable unit of the first memory block. In some embodiments, at a time of receiving the host request, the memory system further includes a space reserve of over-provisioning, and the second memory block is different from the space reserve. In some embodiments, the second memory block is moved offline to join the space reserve.