Title :
Tutorial: Flash-based storage systems modelling, simulation and IO characterisation
Author :
Zertal, Soraya ; Harrison, Peter
Author_Institution :
PRiSM, Univ. de Versailles, Versailles, France
Abstract :
Flash memory becomes an indispensable component among the non-volatile storage devices. Its decreasing cost makes it a challenging candidate for future mass storage systems. Consequently, there is an urgent need for models and simulators to analyse its behaviour and its effects on the global system performance. In this tutorial, we present modelling and simulation tools for Flash memories analysis and methodologies for their workload characterisation. First, a Fluid model with priority classes is used to investigate the response time of Flash accesses, respecting the erase/write/read priorities. Second, The analysis of the impact of different execution modes on the performance of the Flash memory submitted to various IO profiles, is undertaken using a Flash hardware simulator. Finally, a workload analysis methodology is presented that processes data from operation type traces and creates a hidden Markov model (HMM) to represent the workload that generated those traces. The HMM can be used to create representative traces, avoiding the need to repeatedly acquire suitable traces. It can also be used to estimate the transition probabilities and rates of a Markov modulated arrival process directly, for use as input to an analytical performance model of Flash memory.
Keywords :
flash memories; hidden Markov models; random-access storage; HMM; IO characterisation; IO profiles; Markov modulated arrival process; data processing; erase-write-read priorities; execution modes; flash accesses; flash hardware simulator; flash memory performance; flash-based storage system modelling; flash-based storage system simulation; fluid model; hidden Markov model; mass storage systems; nonvolatile storage devices; operation type traces; priority classes; transition probability estimation; workload analysis methodology; workload characterisation; Analytical models; Flash memories; Fluids; Hardware; Hidden Markov models; Markov processes; Time factors; Flash memory; Fluid models; Hidden Markov Models (HMM); IO characterisation; Simulation; Solid State Disks (SSD);
Conference_Titel :
High Performance Computing and Simulation (HPCS), 2013 International Conference on
Conference_Location :
Helsinki
Print_ISBN :
978-1-4799-0836-3
DOI :
10.1109/HPCSim.2013.6641386