NVMeoF Storage Device Integrating Machine Learning for Data Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current NVMeoF devices are not optimized for data-centric applications such as machine learning and data mining, relying on external CPUs and GPUs for data matching and machine learning operations.
Innovation Solution
A data storage device with integrated capabilities for data matching and machine learning, featuring a memory array, host interface, central control unit, preprocessor, and data processing units that can perform search processes, extract features, and return matching data without significant external processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If NVMeoF devices rely on external CPUs and GPUs for data matching and machine learning operations, then device complexity is reduced and ease of manufacture is improved, but processing speed and productivity deteriorate due to data transfer overhead and external processing bottlenecks
Solution Approach 1:
The patent merges the data storage function with data processing capabilities by integrating machine learning processors and data matching engines directly into the NVMeoF device. This allows the storage device to perform machine learning operations and data pattern matching locally without requiring constant communication with external processors, thereby improving processing speed and reducing data transfer overhead.
Solution Approach 2:
The patent adds a new functional dimension to traditional storage devices by incorporating artificial intelligence and machine learning capabilities. This transforms the storage device from a passive data repository into an active processing unit that can perform complex operations such as pattern recognition, classification, and predictive analytics directly on stored data.
2Loss of time
If data matching and machine learning operations are performed on external host processors, then device complexity is reduced, but latency increases due to data transfer between host and storage device
Solution Approach 1:
The NVMeoF device performs data matching and machine learning operations autonomously using integrated processors and algorithms. The device can independently execute search queries, perform pattern matching, and return relevant results without requiring the host system to retrieve all data and perform processing externally. This self-service capability significantly reduces latency by eliminating unnecessary data transfer cycles.
3Adaptability or versatility
If NVMeoF devices perform only basic storage operations, then ease of operation is improved and compatibility is maintained, but adaptability to data-centric applications deteriorates
Solution Approach 1:
The NVMeoF device is designed with multi-functionality to handle both traditional storage operations and advanced data-centric applications. The integrated machine learning processors and configurable algorithms enable the device to perform diverse functions including pattern recognition, classification, clustering, and predictive analytics while maintaining compatibility with standard storage protocols and interfaces.
4Productivity
If data processing capabilities are integrated into the storage device, then productivity and latency are improved, but power consumption increases
Solution Approach 1:
The integrated machine learning processors are designed to perform only the specific data processing tasks required by the storage operations. Rather than continuously running full-scale processing capabilities, the system activates processing power only when needed for data matching, search operations, or machine learning inference, thereby reducing overall power consumption while maintaining high productivity during active operations.
Data Source
AI summary
A data storage device includes a memory array for storing data; a host interface for providing an interface with a host computer running an application; a central control unit configured to receive a command in a submission queue from the application and initiate a search process in response to a search query command; a preprocessor configured to reformat data contained in the search query command and generate a reformatted data; and one or more data processing units configured to extract one or more features from the reformatted data and perform a data operation on the data stored in the memory array in response to the search query command and return matching data from the data stored in the memory array to the application via the host interface.