Storage Device Copy Back via Peer-to-Peer Bypass
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional copy back operations in data storage systems consume significant system resources such as bandwidth, processing cycles, and memory allocation, especially when transferring large data sets, as they typically involve caching data through a storage array controller before transferring it to a destination device.
Innovation Solution
Implementing a peer-to-peer connection between source and destination data storage devices to bypass the storage array controller, allowing direct data transfer using a copy back command that identifies the data set and destination device, thereby reducing the need for intermediate caching and processing overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is transferred through the storage array controller using conventional copy back operations, then data transfer can be managed centrally, but system resources such as bandwidth, processing cycles, and memory allocation are significantly consumed
Solution Approach 1:
The patent extracts the data transfer function from the storage array controller and implements it directly between storage devices through peer-to-peer connections. This removes the bottleneck of controller-mediated transfers and eliminates unnecessary intermediate caching, allowing source and destination devices to communicate directly over the fabric while the controller handles only metadata and control operations.
Solution Approach 2:
The patent introduces a fabric-mediated peer-to-peer connection as an intermediary that enables direct data transfer between storage devices without requiring the storage array controller to handle the actual data movement. The fabric acts as a communication medium that allows devices to establish direct paths, reducing controller processing overhead and bandwidth consumption.
2Ease of operation
If data is cached through the storage array controller before transfer, then data management and control are simplified, but memory allocation and processing overhead increase
Solution Approach 1:
The patent extracts the caching function from the storage array controller memory and implements peer-to-peer direct transfer. Data moves directly from source device memory to destination device memory without being cached in controller memory, eliminating the need for large buffer allocations in the controller while maintaining simplified management through controller-initiated transfer commands.
Solution Approach 2:
The patent enables storage devices to perform their own data transfer operations directly without requiring the controller to cache and manage the data. Source devices read data directly and write it to destination devices, with the controller only initiating and monitoring the transfer, allowing devices to serve themselves rather than relying on controller memory resources.
3Productivity
If peer-to-peer connections are established for direct data transfer, then system resource consumption is reduced, but device complexity increases
Solution Approach 1:
The patent leverages the existing fabric infrastructure to provide peer-to-peer connectivity, using the same communication medium for both traditional controller-mediated operations and direct device-to-device transfers. This multi-functional use of the fabric reduces the need for separate dedicated connection hardware while enabling efficient direct transfers.
4Reliability
If conventional copy back operations are used, then data transfer can be monitored and controlled centrally, but transfer speed and efficiency decrease
Solution Approach 1:
The patent extracts the data path from the controller and creates a direct peer-to-peer connection between source and destination devices. This bypasses the controller processing bottleneck, allowing data to flow directly at line speed while the controller maintains oversight through initiation and completion monitoring, achieving both high speed and reliable control.
Data Source
AI summary
Method and apparatus for managing data in a data storage system. A storage array controller device is coupled to a plurality of storage devices by an external data path, with the storage devices used for non-volatile memory (NVM) storage of user data from a host. A copy back operation is initiated by issuing a copy back transfer command that identifies a selected data set stored in a source device and a unique identifier (ID) value that identifies a destination device. A peer-to-peer connection is established over the external data path in response to the copy back transfer command so that the selected data set is transferred from the source device to the destination device while bypassing the storage array controller device. Normal data transfers can be carried out between the storage array controller and the respective source and destination devices during the copy back operation.