@inproceedings{badam2013better,
author = {Badam, Anirudh and Pai, Vivek S. and Nellans, David W.},
title = {Better Flash Access via Shapeshifting Virtual Memory Pages},
booktitle = {Conference on Timely Results on Operating Systems Principles (TRIOS'13)},
year = {2013},
month = {November},
abstract = {Today, many system designers try to fit the entire data-set of an application in RAM to avoid the cost of accessing magnetic disk. However, for many data-centric applications this is not an option due to the capacity and high $/GB constraints of RAM. As a result, system designers are relying on NAND-Flash to augment RAM. However, rewriting the application to actively tier data between disk, RAM, and flash is a complicated process. Years of research has enabled some applications, namely databases, to tier efficiently, yet there are still large classes of applications that have no built-in tiering and a full application rewrite may take months or years. In this paper, we propose Chameleon, a system to augment RAM with NAND-Flash transparently. Chameleon removes most of the impediments to using flash via virtual memory. Chameleon is the first transparent tiering system to provide low-latency accesses to both RAM and NAND-Flash.
We show that applications using Chameleon, outperform applications using state-of-the-art tiering mechanism by providing more than two orders of magnitude improvement in latency for working sets that can fit in RAM. We also show that Chameleon provides upto 47% latency improvement for out-of-core applications. Finally, we show that Chameleon improves the flash device’s lifetime by upto 8x.},
publisher = {ACM - Association for Computing Machinery},
url = {http://approjects.co.za/?big=en-us/research/publication/better-flash-access-via-shapeshifting-virtual-memory-pages/},
edition = {Conference on Timely Results on Operating Systems Principles (TRIOS'13)},
}