LS-LRU: A Lazy-Split LRU Buffer Replacement Policy for Flash-Based B+-tree Index

Most embedded systems are equipped with flash memory owing to its shock resistance, fast access, and low power consumption. However, some of its distinguishing characteristics, including out-of-place updates, an asymmetric read/write/erase speed, and a limited number of write/erase cycles, make it necessary to reconsider the existing system designs to explore its performance potential. For example, the buffer replacement policy of flash-based systems should not only consider the cache hit ratio, but also the relative heavy write and erase costs that are caused by flushing dirty pages. Most of the recent studies on buffer designs have focused on a Clean-First LRU strategy that evicts clean pages preferentially to reduce the number of writes to flash memory. However, each of them fails to distinguish dirty pages, which may have a different effect on the flash memory. In this paper, we propose a Lazy-Split LRU-based buffer management scheme that not only considers an imbalance of the read and write speeds but also different effects of different dirty pages and frequent changes of the B +

[1]  Sang-Won Lee,et al.  A survey of Flash Translation Layer , 2009, J. Syst. Archit..

[2]  Junseok Park,et al.  DABC-NV: A buffer cache architecture for mobile systems with heterogeneous flash memories , 2012, IEEE Transactions on Consumer Electronics.

[3]  Suman Nath,et al.  Online maintenance of very large random samples on flash storage , 2009, The VLDB Journal.

[4]  Dongkun Shin,et al.  Recently-evicted-first buffer replacement policy for flash storage devices , 2008, IEEE Transactions on Consumer Electronics.

[5]  Peiquan Jin,et al.  CFDC: a flash-aware replacement policy for database buffer management , 2009, DaMoN '09.

[6]  Dong-Ho Lee,et al.  LSF: a new buffer replacement scheme for flash memory-based portable media players , 2013, IEEE Transactions on Consumer Electronics.

[7]  Sang-Won Lee,et al.  A log buffer-based flash translation layer using fully-associative sector translation , 2007, TECS.

[8]  Sang-Won Lee,et al.  Design of flash-based DBMS: an in-page logging approach , 2007, SIGMOD '07.

[9]  Richard E. Neapolitan,et al.  Foundations of Algorithms , 1996 .

[10]  Simonas Saltenis,et al.  FB-tree: a B+-tree for flash-based SSDs , 2011, IDEAS '11.

[11]  Dong-Ho Lee,et al.  An efficient index buffer management scheme for implementing a B-tree on NAND flash memory , 2010, Data Knowl. Eng..

[12]  Shi-Min Hu,et al.  Bridging the information gap between buffer and flash translation layer for flash memory , 2011, IEEE Transactions on Consumer Electronics.

[13]  Xiaofeng Meng,et al.  ACR: An Adaptive Cost-Aware Buffer Replacement Algorithm for Flash Storage Devices , 2010, 2010 Eleventh International Conference on Mobile Data Management.

[14]  Tei-Wei Kuo,et al.  An efficient B-tree layer implementation for flash-memory storage systems , 2007, TECS.

[15]  KuoTei-Wei,et al.  An efficient B-tree layer implementation for flash-memory storage systems , 2007 .

[16]  Tae-Sun Chung,et al.  FlashB-tree: a novel B-tree index scheme for solid state drives , 2011, RACS.

[17]  Joonwon Lee,et al.  CFLRU: a replacement algorithm for flash memory , 2006, CASES '06.