Computational Storage Discovery for In-Place Data Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The inefficiency in processing data stored on storage devices due to the time-consuming movement of data between storage and memory, which can be exacerbated by the need to move data back and forth, is addressed by integrating computational storage units closer to the storage to reduce data transfer time and enable faster processing.
Innovation Solution
The integration of computational storage units within or near storage devices, allowing for in-storage processing, is facilitated by a discovery mechanism that identifies these resources using an application programming interface (API) to enable seamless access and utilization without requiring knowledge of specific interfaces or protocols.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is processed by moving it from storage to memory for the processor, then data processing can be performed, but data transfer time increases and processing efficiency decreases
Solution Approach 1:
The patent combines storage and computation resources into a unified computational storage device. The storage device includes both storage media and processing units (CPUs, GPUs, FPGAs, or ASICs) that can execute data processing operations directly on stored data, eliminating the need to move data between storage and memory systems.
Solution Approach 2:
The patent introduces a discovery service as an intermediary layer between applications and computational storage devices. This service manages the complexity of heterogeneous processing units by providing a unified interface, allowing applications to leverage computational storage capabilities without needing to understand specific hardware interfaces or protocols.
2Speed
If computational storage units are integrated within or near storage devices, then data processing speed increases, but device complexity increases
Solution Approach 1:
The computational storage device is designed to support multiple types of processing units (CPU, GPU, FPGA, ASIC) within a single storage device architecture. This multi-functionality allows the device to handle various data processing tasks while maintaining a standardized interface through the discovery service, thereby managing complexity.
Solution Approach 2:
The discovery service automatically detects and characterizes the computational storage capabilities of connected devices. It performs self-service by identifying available processing units, their types, and their capabilities without requiring manual configuration, thereby simplifying the user's interaction with complex heterogeneous resources.
3Productivity
If data is moved back and forth between storage and memory, then processing operations can be performed, but processing time increases
Solution Approach 1:
By merging storage and computation into a single device, the patent enables processing operations to be performed in-place on stored data. This eliminates the data movement cycle between storage and memory, allowing processing to occur directly where the data resides, thereby reducing total processing time and increasing throughput.
Data Source
AI summary
A device is described. The device may include a connector to connect the device to a component. The device may also include a computational storage unit. A receiver may receive a discovery request from a discovery service, and a transmitter may transmit a discovery response to the discovery service, the discovery response including information about the computational storage unit.