IO Patterns-Aware and Dynamic Scheduling Based Data Placement in Hybrid Storage System

QLC flash offers high reducing the cost of storage sys formance of QLC SSD is not sa write performance. To make matte collection can lead to the serious negative impact on the device life Recently Uesd) method may reliev it comes at the cost of limited $]$ we propose a reinforcement learni strategy for the hybrid storage sys SSDs. First, we investigate the r different flash SSDs. Then, the IO different storage devices with a dy a RF -based method is proposed 1 I/Opatterns of application. Finall is established with considering the features. The real workload is use of the proposed method, and the latency can be reduced by up to other similar methods.