Font Size: a A A

Study On Flash Translation Layer Mechanism For TLC/SLC Dual-mode Flash Memory In Embedded Systems

Posted on:2018-11-12Degree:MasterType:Thesis
Country:ChinaCandidate:L YaoFull Text:PDF
GTID:2348330536468730Subject:Engineering
Abstract/Summary:PDF Full Text Request
Nowadays,NAND flash memory has been widely adopted in embedded systems due to its attractive features,such as non-volatility,low standby power,high density and shock resistance.In particular,TLC(tripe-level cell)flash memory provides much higher density and lower cost than SLC(single-level cell)and MLC(multilevel cell)flash memory.However,TLC flash memory also suffers from longer read/write latency and much lower endurance compared to its counterpart.In order to achieve a balance between performance and capacity,recent advances in TLC flash memory offer the ability of switching flash blocks between TLC mode and SLC mode.Therefore,in this paper,we propose a workload-aware FTL(flash translation layer),named Balloon-FTL,for TLC/SLC dual-mode flash memory in embedded systems,to improve performance and lifespan of the system.Balloon-FTL consists of two modules: a workload identifier and a hybrid mapping mechanism.Workload identifier is proposed to dynamically determine the TLC/SLC capacity based on workload of different applications.Hybrid mapping mechanism is proposed for mapping pages at distinct granularity in TLC/SLC dual-mode flash memory according to different I/O workloads.Evaluation results show that Balloon-FTL can effectively improve the system performance and lifespan.This work makes the following contributions:First,we propose a workload identifier module to dynamically allocate TLC/SLC capacity based on different application workloads for achieving a balanced write distribution in flash memory,to improve the lifespan of the whole system with low memory access cost.Second,we propose a hybrid FTL to map pages at distinct granularity in TLC/SLC dual-mode flash memory according to different I/O workloads.To achieve this,a page-level mapping is adopted to handle random and hot sequential requests,while a block-level mapping is adopted to handle cold sequential requests.Third,we develop a simulator to evaluate the Balloon-FTL for TLC/SLC dual-mode Flash memory in embedded systems and conduct a series of experiments to evaluate its effectiveness with a set of realistic I/O traces.
Keywords/Search Tags:flash translation layer, TLC/SLC dual-mode flash memory, workload aware, genetic algorithm
PDF Full Text Request
Related items