Detailed Information

Cited 0 time in webofscience Cited 0 time in scopus
Metadata Downloads

RMSS: An Efficient Recovery Management Scheme on NAND Flash Memory based Solid State Disk

Authors
Lee, Hyun-SeobPark, SangwonLee, Dong-Ho
Issue Date
Feb-2013
Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
Keywords
Flash Memory; FTL; Address Translation; Recovery
Citation
IEEE TRANSACTIONS ON CONSUMER ELECTRONICS, v.59, no.1, pp 107 - 112
Pages
6
Indexed
SCI
SCIE
SCOPUS
Journal Title
IEEE TRANSACTIONS ON CONSUMER ELECTRONICS
Volume
59
Number
1
Start Page
107
End Page
112
URI
https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/28866
DOI
10.1109/TCE.2013.6490248
ISSN
0098-3063
1558-4127
Abstract
In many consumer electronics such as digital camcorders, notebooks, and tablet PCs, hard disk drive (HDD) has been replaced with NAND flash memory based solid state disk (SSD) because of its fast speed and low power consumption. However, since SSD inherits the limitations of NAND flash memory such as erase-before-write architecture and asymmetric read, write, and erase speeds, it may result in severe performance degradation to implement a B-tree on SSD. To address these problems, several methods exploiting the buffer have been proposed so far. However, they have faced with the recovery problem because all index data in the buffer are lost when a sudden power-off occurs. In this paper, we introduce a method called RMSS (recovery management scheme on SSD) that supports an efficient recovery mechanism when a B-tree is built on SSD. Since RMSS flushes all index data and creates a checkpoint whenever updating the root node, it can stably restore the index structure into up-to-date and consistent state. Consequently, RMSS efficiently implements a B-tree on SSD by using a buffer, and also recovers the original B-tree when a power failure occurs. We show the performance of RMSS on SSD through various experiments(1).
Files in This Item
Go to Link
Appears in
Collections
COLLEGE OF COMPUTING > DEPARTMENT OF ARTIFICIAL INTELLIGENCE > 1. Journal Articles

qrcode

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.

Related Researcher

Researcher Lee, Dong Ho photo

Lee, Dong Ho
ERICA 소프트웨어융합대학 (DEPARTMENT OF ARTIFICIAL INTELLIGENCE)
Read more

Altmetrics

Total Views & Downloads

BROWSE