NAND Logical Partitioning for Dynamic Resource Pooling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional memory partitioning methods in NAND devices require separate physical partitions, which can lead to resource inefficiencies and limitations in managing memory cells, as they cannot dynamically reconfigure or allocate resources across partitions, potentially resulting in unusable partitions due to uneven wear and tear.
Innovation Solution
Implementing a NAND-level logical partitioning system where the memory controller manages memory as a shared pool, allowing for dynamic allocation and reconfiguration of memory cells across partitions, and translating host requests to allocate physical resources efficiently, ensuring service level guarantees without the need for separate physical partitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate physical partitions are used in traditional memory devices, then partition isolation and security are improved, but resource utilization deteriorates due to inability to dynamically reconfigure memory cells
Solution Approach 1:
The memory device is segmented into multiple logical partitions that are software-defined rather than physically separated. The controller divides the memory space into logical partitions (e.g., LUNs) that can be dynamically created, modified, and deleted without physical reconfiguration, maintaining isolation through logical addressing while enabling flexible resource allocation.
Solution Approach 2:
The system implements dynamic partitioning where memory cells can be reallocated between partitions based on demand. The controller manages a pool of memory cells that can be dynamically assigned to different logical partitions, allowing memory capacity to be flexibly adjusted and reconfigured without physical changes, thus improving resource utilization while maintaining partition isolation.
2Reliability
If separate physical partitions are used, then partition security is improved, but adaptability deteriorates due to fixed physical boundaries
Solution Approach 1:
The system implements dynamic partitioning where memory cells can be reallocated between partitions based on demand. The controller manages a pool of memory cells that can be dynamically assigned to different logical partitions, allowing memory capacity to be flexibly adjusted and reconfigured without physical changes, thus improving resource utilization.
Solution Approach 2:
The same physical memory pool serves multiple logical partitions simultaneously. The controller enables a single physical memory resource to be shared across multiple logical partitions with different access permissions, security levels, and capacity requirements, providing universal access control and multi-functional usage from a single physical substrate.
3Ease of operation
If memory cells are statically allocated to partitions, then management simplicity is improved, but resource efficiency deteriorates due to uneven wear and unused capacity
Solution Approach 1:
The system implements dynamic partitioning where memory cells can be reallocated between partitions based on demand. The controller manages a pool of memory cells that can be dynamically assigned to different logical partitions, allowing memory capacity to be flexibly adjusted and reconfigured without physical changes, thus improving resource utilization.
Solution Approach 2:
The system changes the parameter of partition boundaries from fixed physical locations to flexible logical addresses. The controller dynamically adjusts partition parameters (capacity, allocation, mapping) based on wear levels, usage patterns, and performance requirements, enabling simple management interfaces while achieving high resource efficiency through parameter optimization.
4Reliability
If physical partitions are created, then partition independence is improved, but device complexity increases due to separate management of each partition
Solution Approach 1:
The system merges multiple physical partition management functions into a single unified logical partitioning layer. The controller combines memory cell management, address translation, and partition isolation into one integrated logical partitioning system, reducing device complexity by eliminating the need for separate physical management structures while maintaining partition independence through software-defined boundaries.
Data Source
AI summary
Disclosed in some examples are methods, systems, machine-readable mediums, and NAND devices which create logical partitions when requested to create a physical partition. The controller on the NAND mimics the creation of the physical partition to the host device that requested the physical partition. Thus, the host device sees the logical partition as a physical partition. Despite this, the NAND does not incur the memory storage expense of creating a separate partition, and additionally the NAND can borrow cells for overprovisioning from another partition. In these examples, a host device operating system believes that a physical partition has been created, but the NAND manages the memory as a contiguous pool of resources. Thus, a logical partition is created at the NAND memory controller level—as opposed to at the operating system level.