Copy Offload via Persistent Point-in-Time Image
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage methods for copy offload operations face inefficiencies due to the need for host-centered data transfer, which consumes CPU resources and network bandwidth, and existing solutions like write tracking or making additional copies introduce significant overhead.
Innovation Solution
Implementing persistent point-in-time images using data pointers that create a snapshot of data without duplicating data blocks, allowing for efficient copy offload operations without the need for write tracking or additional storage, enabling direct data transfer between storage devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional host-centered copy process is used, then data can be transferred between storage devices, but CPU resources and network bandwidth are consumed
Solution Approach 1:
The patent extracts the data transfer operation from the host system and relocates it to the storage device. The host only sends control commands (token creation, copy initiation), while the actual data copying is performed by the storage device's internal copy engine, eliminating the need for host CPU involvement in the data transfer process.
Solution Approach 2:
The storage device performs self-service by autonomously copying data between its internal storage locations. The device receives high-level commands from the host but independently manages the entire copy operation, including data retrieval, validation, and transfer, without requiring host intervention or resource consumption.
2Reliability
If write tracking is implemented to maintain token validity, then data integrity is preserved, but system overhead increases
Solution Approach 1:
The patent performs preliminary action by creating a consistent snapshot of the data at the time of token creation. This snapshot captures the data state before any subsequent writes, ensuring token validity without requiring continuous write tracking. The snapshot is created once at token issuance, eliminating the need for ongoing monitoring.
Solution Approach 2:
The patent extracts the data consistency guarantee from the write tracking mechanism and embeds it directly in the token structure. The token carries snapshot metadata that inherently validates data consistency, removing the need for separate write tracking operations and reducing system overhead.
3Reliability
If additional copies of data are made for token representation, then data consistency is maintained, but storage space consumption increases
Solution Approach 1:
The patent uses a metadata copy approach rather than full data copying. The token contains a compact representation of the data snapshot, including pointers or references to the actual data blocks. This allows multiple tokens to reference the same underlying data without duplicating the data itself, maintaining consistency while minimizing storage space consumption.
Solution Approach 2:
The patent implements a nested structure where the token (smaller object) contains references to the actual data blocks (larger objects). The token acts as a container that points to the real data, allowing the system to maintain data consistency through the token while avoiding duplication of the large data blocks. This nested approach efficiently uses storage space.
Data Source
AI summary
Methods and apparatuses for performing simplified copy offload operations are provided. In one exemplary embodiment, a method includes creating a persistent point-in-time image of a data set of a source storage device in response to receiving a token request for a copy offload operation from a host. Creating the persistent point-in-time image includes generating pointers to data blocks of the data set, but does not include duplicating the data blocks. The method also includes transmitting a token that is a representation of the persistent point-in-time image to the host in response to the request.