uFlip: Understanding Flash IO Patterns

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskningfagfællebedømt

Standard

uFlip: Understanding Flash IO Patterns. / Bouganim, Luc; Jonsson, Bjørn; Bonnet, Philippe.

CIDR. 2009. s. 1-12.

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskningfagfællebedømt

Harvard

Bouganim, L, Jonsson, B & Bonnet, P 2009, uFlip: Understanding Flash IO Patterns. i CIDR. s. 1-12, The Biennial Conference on Innovative Data systems Research (CIDR '09), Asilomar, CA, USA, 04/01/2009.

APA

Bouganim, L., Jonsson, B., & Bonnet, P. (2009). uFlip: Understanding Flash IO Patterns. I CIDR (s. 1-12)

Vancouver

Bouganim L, Jonsson B, Bonnet P. uFlip: Understanding Flash IO Patterns. I CIDR. 2009. s. 1-12

Author

Bouganim, Luc ; Jonsson, Bjørn ; Bonnet, Philippe. / uFlip: Understanding Flash IO Patterns. CIDR. 2009. s. 1-12

Bibtex

@inproceedings{fb139cd0f5c211ddbf70000ea68e967b,
title = "uFlip: Understanding Flash IO Patterns",
abstract = " Does the advent of flash devices constitute a radical change forsecondary storage? How should database systems adapt to thisnew form of secondary storage? Before we can answer thesequestions, we need to fully understand the performancecharacteristics of flash devices. More specifically, we want toestablish what kind of IOs should be favored (or avoided) whendesigning algorithms and architectures for flash-based systems. Inthis paper, we focus on flash IO patterns, that capture relevantdistribution of IOs in time and space, and our goal is to quantifytheir performance. We define uFLIP, a benchmark for measuringthe response time of flash IO patterns. We also present abenchmarking methodology which takes into account the particularcharacteristics of flash devices. Finally, we present the resultsobtained by measuring eleven flash devices, and derive a set ofdesign hints that should drive the development of flash-basedsystems on current devices.",
author = "Luc Bouganim and Bj{\o}rn Jonsson and Philippe Bonnet",
year = "2009",
language = "English",
pages = "1--12",
booktitle = "CIDR",
note = "null ; Conference date: 04-01-2009 Through 07-01-2009",

}

RIS

TY - GEN

T1 - uFlip: Understanding Flash IO Patterns

AU - Bouganim, Luc

AU - Jonsson, Bjørn

AU - Bonnet, Philippe

N1 - Conference code: 4

PY - 2009

Y1 - 2009

N2 -  Does the advent of flash devices constitute a radical change forsecondary storage? How should database systems adapt to thisnew form of secondary storage? Before we can answer thesequestions, we need to fully understand the performancecharacteristics of flash devices. More specifically, we want toestablish what kind of IOs should be favored (or avoided) whendesigning algorithms and architectures for flash-based systems. Inthis paper, we focus on flash IO patterns, that capture relevantdistribution of IOs in time and space, and our goal is to quantifytheir performance. We define uFLIP, a benchmark for measuringthe response time of flash IO patterns. We also present abenchmarking methodology which takes into account the particularcharacteristics of flash devices. Finally, we present the resultsobtained by measuring eleven flash devices, and derive a set ofdesign hints that should drive the development of flash-basedsystems on current devices.

AB -  Does the advent of flash devices constitute a radical change forsecondary storage? How should database systems adapt to thisnew form of secondary storage? Before we can answer thesequestions, we need to fully understand the performancecharacteristics of flash devices. More specifically, we want toestablish what kind of IOs should be favored (or avoided) whendesigning algorithms and architectures for flash-based systems. Inthis paper, we focus on flash IO patterns, that capture relevantdistribution of IOs in time and space, and our goal is to quantifytheir performance. We define uFLIP, a benchmark for measuringthe response time of flash IO patterns. We also present abenchmarking methodology which takes into account the particularcharacteristics of flash devices. Finally, we present the resultsobtained by measuring eleven flash devices, and derive a set ofdesign hints that should drive the development of flash-basedsystems on current devices.

M3 - Article in proceedings

SP - 1

EP - 12

BT - CIDR

Y2 - 4 January 2009 through 7 January 2009

ER -

ID: 10186082