17819007. STORAGE SYSTEM simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
Contents
STORAGE SYSTEM
Organization Name
Inventor(s)
Byeong Hui Kim of Hwaseong-si (KR)
Sung Min Jang of Hwaseong-si (KR)
STORAGE SYSTEM - A simplified explanation of the abstract
This abstract first appeared for US patent application 17819007 titled 'STORAGE SYSTEM
Simplified Explanation
The abstract describes a storage system connected to a machine learning model embedded device that uses a trained simulation model to optimize Quality of Service (QoS) data. The simulation model is trained using a loss function that compares predicted QoS data with real QoS data. The machine learning model embedded device includes a map table to store optimized parameters for maximizing QoS relative to different workloads. The storage controller can access the storage device and perform workloads based on the optimized parameters selected by the trained simulation model.
- A storage system is connected to a machine learning model embedded device.
- The machine learning model uses a trained simulation model to optimize QoS data.
- The simulation model is trained using a loss function that compares predicted and real QoS data.
- The simulation model can search for a workload to maximize QoS using a fixed parameter or vice versa.
- The machine learning model embedded device includes a map table to store optimized parameters for each workload.
- The storage controller can access the storage device and perform workloads based on the optimized parameters selected by the simulation model.
Potential Applications
- Storage systems with optimized QoS for different workloads.
- Improved performance and efficiency in storage operations.
- Enhanced user experience with faster and more reliable storage access.
Problems Solved
- Finding the optimal parameters for maximizing QoS in storage systems.
- Improving the performance and efficiency of storage operations.
- Addressing the challenge of workload optimization in storage systems.
Benefits
- Increased Quality of Service (QoS) in storage operations.
- Improved performance and efficiency of storage systems.
- Enhanced user satisfaction with faster and more reliable storage access.
Original Abstract Submitted
A storage system (e.g., a storage device, a storage controller, etc.) may be connected to a machine learning model embedded device that includes a trained simulation model. The simulation model may be trained based on a loss function, where the loss function is based on a comparison of predicted QoS data and real QoS data. For example, the simulation model may be trained to search for a workload for increasing (e.g., maximizing) the QoS using a fixed parameter or, conversely, to search for a parameter for maximizing the QoS using a fixed workload. Further, the machine learning model embedded device may include a map table storing the optimized parameters calculated by the trained simulation model so as to maximize QoS relative to each workload. Accordingly, a storage controller may be configured to access a storage device and perform workloads based on optimized parameters selected by the trained simulation model.
