Network I/O Device Cache Buffering for Storage Data Staging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing network data capacities and lower access latencies in newer storage technologies, such as solid state drives, make the use of system memory for routing data between network and storage subsystems less efficient, leading to performance degradation due to overhead and latencies.
Innovation Solution
Implementing a network I/O device with circuitry capable of executing modules to manage data movement between network and storage subsystems using buffers in a cache, allowing for efficient data staging and control information exchange without relying on system memory, and integrating network and storage subsystems to facilitate direct data access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If system memory is used to route data between network and storage subsystems, then data movement capability is provided, but throughput performance degrades due to overhead and latencies
Solution Approach 1:
The patent extracts the data staging function from system memory and relocates it to buffers within the cache of the network I/O device. This removes the bottleneck of system memory access from the critical data path between network and storage subsystems, eliminating the overhead and latencies associated with system memory routing while maintaining the necessary data buffering capability.
Solution Approach 2:
The patent introduces buffers in the cache of the network I/O device as an intermediary structure between the network subsystem and storage subsystem. These buffers serve as a local staging area that eliminates the need for system memory involvement in data routing, thereby reducing latency and overhead while enabling direct data movement between subsystems.
2Productivity
If system memory is used for data routing between subsystems, then data transfer capability is enabled, but overhead increases reducing efficiency
Solution Approach 1:
The patent merges the data staging function with the cache memory of the network I/O device. By combining the buffering capability directly into the network I/O device's cache structure, the system eliminates the separate system memory routing step, reducing overhead and simplifying the data path while maintaining full data transfer capability between network and storage subsystems.
Data Source
AI summary
Examples are disclosed for moving data between a network input/output (I/O) device and a storage subsystem and/or storage device. In some examples, a network I/O device coupled to a host device may receive a data frame including a request to access a storage subsystem or storage device. The storage subsystem and/or storage device may be located with the network I/O device or separately coupled to the host device through a storage controller. One or more buffers maintained in a cache for processor circuitry may be used to exchange control information or stage data associated with the data frame to avoid or eliminate use of system memory to move data to or from the storage subsystem and/or storage device. Other examples are described and claimed.