SSD Copy Mode Using Source L2P Data to Cut Write Amplification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage device-to-device copy processes, particularly in virtual storage pools or data centers, suffer from increased write amplification and performance overhead due to the generation of temporary parity and logical-to-physical address translation information during the transfer of data from old to new SSDs, consuming significant RAM and CPU resources.
Innovation Solution
A new data copy mode where valid blocks from a source SSD are directly copied to a target SSD without generating parity or logical-to-physical address translation information, using the source's existing information and allowing the host to manage retries and block relocations, reducing the need for temporary parity and L2P entry generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If standard read/write commands are used to copy data from source SSD to target SSD, then data transfer can be performed, but write amplification increases and performance deteriorates due to generation of parity and L2P translation information
Solution Approach 1:
The host system performs preliminary actions by reading logical-to-physical address translation information from the source SSD before the copy operation. This pre-fetched translation data is then provided to the target SSD during the copy process, eliminating the need for the target SSD to generate its own L2P translation information and reducing write amplification.
Solution Approach 2:
The host system acts as an intermediary that mediates between the source SSD and target SSD. It reads translation information from the source and provides it to the target, enabling direct data copying without requiring the target SSD to perform complex translation operations, thus reducing write amplification and improving transfer speed.
2Reliability
If target SSD generates parity information and L2P translation information during copy, then data integrity and address translation are ensured, but RAM and CPU resources are consumed
Solution Approach 1:
The host system performs preliminary action by reading and caching the logical-to-physical address translation information from the source SSD before initiating the copy operation. This pre-fetched translation data is then supplied to the target SSD, eliminating the need for the target SSD to generate its own translation information and reducing RAM and CPU resource consumption.
Solution Approach 2:
Instead of having the target SSD regenerate translation information, the system copies the existing translation information from the source SSD directly to the target SSD. This copying approach preserves data integrity while significantly reducing the computational resources required for translation operations.
3Productivity
If full data copy with parity and translation generation is performed, then data can be transferred to new SSD, but copy time increases
Solution Approach 1:
The host system reads and prepares the logical-to-physical address translation information from the source SSD before the copy operation begins. This preliminary preparation of translation data eliminates the time-consuming generation process at the target SSD during the actual copy operation, thereby reducing overall copy time and improving copy speed.
Solution Approach 2:
The system extracts the logical-to-physical address translation information from the source SSD and separates it from the data copy process. By extracting this translation data beforehand, the target SSD can focus solely on receiving and writing the actual data, significantly reducing the time required for the copy operation.
Data Source
AI summary
A data storage device and method for efficient data-storage-device-to-data-storage-device copying are disclosed. In one embodiment, a data storage device is provided comprising a memory and one or more processors. The one or more processors, individually or in combination, are configured to: receive, from a host, a command to place the data storage device in a data-storage-device-to-data-storage-device copy mode; receive, from a source data storage device via the host, a plurality of blocks of data and logical-to-physical address translation information for the plurality of blocks of data; and write, in the memory, the plurality of blocks of data and the logical-to-physical address translation information for the plurality of blocks of data without creating parity information or new logical-to-physical address translation information for the plurality of blocks of data. Other embodiments are provided.