网站开发项目组团队,相亲网站认识的可以做朋友,在线做效果图的网站,云南seo刷关键词排名优化自己想做的#xff0c;刘爱贵在2010年就做完了(⊙ω⊙) http://blog.csdn.net/liuaigui/article/details/5521024 通过综合使用多种标准文件系统Benchmarks对Ext3, Ext4, Reiserfs, XFS, JFS, Reiser4的性能测试对比#xff0c;对不同应用选择合适的文件系统给出以下方案刘爱贵在2010年就做完了(⊙ω⊙) http://blog.csdn.net/liuaigui/article/details/5521024 通过综合使用多种标准文件系统Benchmarks对Ext3, Ext4, Reiserfs, XFS, JFS, Reiser4的性能测试对比对不同应用选择合适的文件系统给出以下方案供大家参考。文件系统性能测试数据见附表。 1、大量小文件LOSF, Lost of small filesI/O应用(如小图片) Reiserfs(首选), Ext4文件系统适合这类负载特征IO调度算法选择deadlineblock size 4096, ext4关闭日志功能。 reiserfs mount参数-o defaults, async, noatime, nodiratime, notail, datawriteback ext4 mount参数-o defaults, async, noatime, nodiratime, datawriteback, barrier0 关闭ext4日志tune2fs -O^has_journal /dev/sdXX 2、大文件I/O应用(如视频下载、流媒体) EXT4文件系统适合此类负载特征IO调度算法选择anticipatory, block size 4096, 关闭日志功能启用extent(default)。 mount参数-o defaults, async, noatime, nodiratime, datawriteback, barrier0 关闭ext4日志tune2fs -O^has_journal /dev/sdXX 3、SSD文件系统选择 EXT4/Reiserfs可以作为SSD文件系统但未对SSD做优化不能充分发挥SSD性能并影响SSD使用时间。 Btrfs对SSD作了优化mount通过参数启用。但Btrfs仍处于实验阶段生产环境谨慎使用。 JFFS2/Nilfs2/YAFFS是常用的flash file system在嵌入式环境广泛应用建议使用。性能目前还未作测试评估。 简单分析一下选择Reiserfs和ext4文件系统的原因 1、Reiserfs 大量小文件访问衡量指标是IOPS文件系统性能瓶颈在于文件元数据操作、目录操作、数据寻址。reiserfs对小文件作了优化并使用B tree组织数据加速了数据寻址大大降低了open/create/delete/close等系统调用开销。mount时指定noatime, nodiratime, notail减少不必要的inode操作notail关闭tail package功能以空间换取更高性能。因此对于随机的小I/O读写reiserfs是很好的选择。 2、Ext4 大文件顺序访问衡量指标是IO吞吐量文件系统性能瓶颈在于数据块布局(layout)、数据寻址。Ext4对ext3主要作了两方面的优化: 一是inode预分配。这使得inode具有很好的局部性特征同一目录文件inode尽量放在一起加速了目录寻址与操作性能。因此在小文件应用方面也具有很好的性能表现。 二是extent/delay/multi的数据块分配策略。这些策略使得大文件的数据块保持连续存储在磁盘上数据寻址次数大大减少显著提高I/O吞吐量。 因此对于顺序大I/O读写EXT4是很好的选择。另外XFS性能在大文件方面也相当不错。 附表文件系统性能测试数据 Testing Host CPUGenuine Intel(R) 2.13GHz 4 cores MEM4GB DISKSAS 146.8GB * 4 /dev/sdbExt3disable journaldefaults,async,noatime,nodiratime /dev/sdcExt4disable journaldefaults,async,noatime,nodiratime,datawriteback,barrier0 /dev/sddXFShas_journaldefaults,async,noatime,nodiratime,barrier0 /dev/sdeReiser4has_journaldefaults,async,noatime,nodiratime /dev/sdereiserfshas_journaldefaults,async,noatime,nodiratime,notail,datawriteback /dev/sdeJFShas_journaldefaults,async,noatime,nodiratime /dev/sdebtrfshas_journaldefaults.async.noatime,nodiratime,datawriteback NO.Test PointsExt3Ext4ReiserfsXFSJFSReiser4Btrfs1filesystem creation s37.38122.3333.7441.7861.8582.758 2filesystem mounting s0.0240.0390.2990.0790.0683.524 3filesystem unmounting s0.0110.020.0740.0250.0480.037 4copy a large file (4GB) s35.48637.22730.60543.54730.3242.17 5recopy the large file s56.91957.52659.89359.31555.65457.925 6remove the large file s3.1260.2952.1780.0160.0184.137 7copy linux kernel archive.tar s0.1030.1430.5740.0940.5520.564 8extract files from kernel.tar s42.6784439.68157.37470.99347.538 9create kernel.tar from linux source s32.73339.14539.30539.16737.9243.828 10compile linux kernel s561.834566.474557.429571.3551773.143828.028 11copy kernel file tree s190.701197.191194.864217.355255.777185.406 12recopy kernel file tree s66.272170.643100.67143.517194.89879.247 13remove kernel file tree s13.7193.7833.66150.07292.06833.2 14list all files for kernel source(tree) s15.96115.53920.71715.81829.13915.611 15file search in kernel source (find) s0.1190.90.3312.5150.7915.913 16create 1 million files s1000.969986.712940.2011128.2441111.5092462.65 17recopy the 1 million files s4783.069595.82859.252972.6711005.0571734.48 18remove the 1 million files s3257.384396.3274.917735.0482395.4171165.314 19dd read IOPS (bs1KB,4GB) s26.79527.25726.03626.15825.94826.485 20dd write IOPS (bs1KB,4GB) s27.00626.46132.57726.16429.203126.906 21dd read BW (bs1MB,4GB) s26.78423.55225.99926.23328.63526.541 22dd write BW (bs1MB,4GB) s29.42926.12330.6424.68331.44631.925 Postmark (postmark.conf) 23transaction /s333384384113166416 24read MB/s27.9820.4224.379.6910.6430.22 25write MB/s87.3263.7276.0630.2333.2194.31 Randomio (randomio bigfile 10 .25 .01 2048 60 1)(X表示Randomio不支持该文件系统) 26create 4GB file MB/s16716616416815651.3 27random io/s372.1382.3378.8521.5382.7X 28random read latency ms26.926.226.613.226.3X 29random write latency ms26.52625.937.125.6X 30random read latency std dev ms17.916.716.613.116.3X 31random write latency std dev ms15.814.914.538.814.3X Bonnie (-d /cacheX -s 7920 -x 1 -u root)(表示未获得准确测试结果) 32read KB/s155161150029160919159242155811144079 33write KB/s155150150635142801159393139770107923 34seek /s812.9816.7816806.4730.7775.9 35sequential create /s7374()29088710136610031 36sequential delete /s()()251606354703123 37random create /s7558()281787455095015 38random delete /s25292()247235592873365 IOZone (-Razc -s 8192m -y 4k|1m -q 4k|1m -i 0 -i 1 -i 2 -b iozone.xls) 39sequential read with 4KB KB/s 40sequential write with 4KB KB/s (测试用时太长暂未完成) 41random read with 4KB KB/s 42random write with 4KB KB/s 43sequential read with 1MB KB/s156790153432160142159591161785143197 44sequential write with 1MB KB/s155928156378131246162226143348138672 45random read with 1MB KB/s1010049806010284010122510110095864 46random write with 1MB KB/s1105371022459792410714011973397941 FileBench (load) 47randomread 48randomwrite 49varmail 49iops /s1292.331921.6581276.035710.653692.282772.895 BW MB/s4.26.34.32.42.32.6 webproxy 50iops /s14868.18916379.94214984.046156.1592830.0368705.437 51BW MB/s55.560.85625.112.334.7 videoserver iops /s395.147392.849391.85391.65391.74391.048 BW MB/s97.5227.996.7227.696.796.5 转载于:https://www.cnblogs.com/tcicy/p/8458359.html
