Web server workload characterization: the search for invariants
Proceedings of the 1996 ACM SIGMETRICS international conference on Measurement and modeling of computer systems
A large-scale study of file-system contents
SIGMETRICS '99 Proceedings of the 1999 ACM SIGMETRICS international conference on Measurement and modeling of computer systems
FlashCache: a NAND flash memory file cache for low power web servers
CASES '06 Proceedings of the 2006 international conference on Compilers, architecture and synthesis for embedded systems
A comparison of file system workloads
ATEC '00 Proceedings of the annual conference on USENIX Annual Technical Conference
A log buffer-based flash translation layer using fully-associative sector translation
ACM Transactions on Embedded Computing Systems (TECS)
A space-efficient flash translation layer for CompactFlash systems
IEEE Transactions on Consumer Electronics
| Hi-index | 0.00 |
To support conventional file systems such as FAT, the NAND flash memory needs the FTL to provide transparent block device emulation to the file system. However, the log-block FTL, the most popularly used FTL, suffers from the performance decline in workloads that contain many random writes. When the file system is not aware of the NAND flash, it often allocates files fragmented in the NAND flash, and this causes many random writes. Consequently, to improve the performance of the FTL, the file system must allocate files defragmented in the NAND flash. In this paper, we propose a NAND flash-aware cluster allocation method for FAT file system, named FECA (Flash-aware Extension-based Cluster Allocation). FECA exploits the following two observations. The first one is that the effort to defragment small-sized files may not improve the performance at all times. The second one is that there is a very strong correlation between the size and the filename extension of files in most cases. Based on those observations, FECA predicts sizes of files by using their extensions and determines the allocation policy for them. To evaluate the effectiveness of FECA, we devise two defragmentation metrics considering the features of the NAND flash. We prove that FECA outperforms previous methods in terms of both metrics through extensive experiments. The results show that FECA improves the performance by 10 % and reduces the garbage collection frequency up to 35% compared to the previous methods.