Staging Sub-Drive SSD Garbage Collection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Solid state drives (SSDs) face performance burdens due to the garbage collection process, which is necessary because of the mismatch between write and erase granularities in block-oriented storage devices, leading to significant write amplification and data management overheads, especially when 'hot' and 'cold' data are mixed in the same blocks.
Innovation Solution
Implementing a system with a staging sub-drive that receives all initial host data writes and only moves data during garbage collection to free space, while selecting sub-drives for garbage collection based on temperature ranges and overprovisioning thresholds, thereby relocating valid data pieces to sub-drives matching their associated temperature ranges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If garbage collection is performed to free space in non-volatile memory, then new data can be written, but processing resources are burdened and performance is reduced
Solution Approach 1:
The patent segments the non-volatile memory into multiple sub-drives (first sub-drive, second sub-drive, third sub-drive) with different functions. The first sub-drive accepts all host data writes, the second sub-drive stores cold data, and the third sub-drive stores hot data. This segmentation allows garbage collection to be performed independently in each sub-drive, reducing the overall processing burden compared to managing a single unified space.
Solution Approach 2:
The first sub-drive acts as an intermediary staging area between the host and the other sub-drives. All incoming data is written to the first sub-drive, which then serves as the source for garbage collection operations that move data to the second and third sub-drives. This intermediary structure simplifies the garbage collection process by providing a dedicated source sub-drive.
2Quantity of substance
If data of different temperatures are mixed in the same blocks, then storage space is utilized, but write amplification increases significantly
Solution Approach 1:
The patent applies local quality by creating sub-drives with specialized functions tailored to specific data temperature characteristics. The second sub-drive is optimized for cold data (low write frequency) while the third sub-drive is optimized for hot data (high write frequency). By placing data with similar access patterns in the same physical location (sub-drive), the system reduces write amplification because garbage collection can target entire sub-drives rather than scattered blocks across the whole memory.
Solution Approach 2:
The patent introduces a new dimension of organization by adding the temperature-based classification dimension to the traditional storage structure. Instead of a single-dimensional flat storage space, the system creates a multi-dimensional structure where data is first organized by temperature (cold vs. hot) and then managed separately. This dimensional change enables more efficient garbage collection by allowing the system to operate on temperature-specific sub-drives rather than mixing all data together.
3Ease of operation
If a staging sub-drive is used to receive all host data writes, then data management is simplified, but additional data movement operations are required
Solution Approach 1:
The system performs preliminary action by writing all incoming host data to the first sub-drive (staging area) before final placement in the second or third sub-drives. This preliminary write to a standardized location simplifies the data management interface with the host, as all writes go to the same place regardless of data temperature. The temperature classification and final placement happen asynchronously during garbage collection operations, separating the simple write path from the more complex classification logic.
Solution Approach 2:
The staging sub-drive structure enables self-service by allowing the system to automatically classify and move data from the first sub-drive to the appropriate second or third sub-drive during garbage collection operations. The temperature classification is performed automatically based on write patterns, and data is moved autonomously without requiring complex real-time decision-making during host write operations. This self-service approach simplifies the write path while handling classification in the background.
Data Source
AI summary
A system and method is disclosed for managing data in a non-volatile memory. The system may include a non-volatile memory having multiple non-volatile memory sub-drives, including a staging sub-drive to receive all data from a host and a plurality of other sub-drives each associated with a respective data temperature range. A controller of the memory system is configured to route all incoming host data only to the staging sub-drive and during garbage collection each individual piece of valid data from a selected source block in a selected source sub-drive is routed to a respective one of the sub-drives. The method may include only routing host data to the staging sub-drive and only relocating valid data to sub-drives other than the staging sub-drive based on a determined temperature of valid data and a unique temperature range associated with sub-drives other than the staging sub-drive in the non-volatile memory system.