软RAID0：花式折腾阿里云

0b0000：起因

最近在琢磨块链条上的东西，打算在自己的阿里云上也折腾折腾；上来就遇到了一个麻烦：磁盘IO有点小，同步速度感觉不太友善，于是就想琢磨琢磨怎么加速。

打开阿里的云盘购买界面，80G的SSD，1小时价格0.112，一年下来980多，顿时就对这个看钱的世界绝望了；冷静分析了一番，最后决定买2个40G的”高效云盘”来加做个RAID0试试（2个40G高效云盘的价格是0.04每小时，折算下来每年大概350来块，瞬间感觉自己节约了好几个亿）。

0b0001：动手开干，创建RAID设备

购买了云盘之后将其挂载到VPS上，让我们先来验证下盘是否正常，运行 ls -l /dev | grep vd ，输出

然后运行 mdadm --examine /dev/vd[b-c]来测试下（如果提示命令不存在，先 yum install mdadm或 apt-get install mdadm安装该工具），此时应该输出

紧接给这俩盘创建逻辑分区，运行 fdisk /dev/vdb，按下面步操作

1. 输入“n”，创建一个新的分区
2. 输入“p”，代表创建的是基本分区
3. 接下来会询问分区编号和起止扇区，默认按回车即可
4. 输入“p”来查看已创建的分区，确保下面的列表已经出现“/dev/vdb1”
5. 输入“t”，设置分区的类型
6. 输入“fd”，代表将该分区设置为’Linux raid autodetect’类型
7. 输入“w”，保存更改

当然，vdc也要进行相同的操作

这时再运行 mdadm --examine /dev/vd[b-c]此时应该输出类似信息：

然后运行 mdadm --create /dev/md0 --level=stripe --raid-devices=2 /dev/vd[b-c]1或者 mdadm -C /dev/md0 -l raid0 -n 2 /dev/vd[b-c]1 此时应该输出类似信息：

此时 cat /proc/mdstat的话已经可以看到相应的raid信息：

如果再保险点的话，分别再运行一下 mdadm -E /dev/vd[b-c]1和 mdadm --detail /dev/md0来再次确认相关信息

确认Ok后运行 mdadm -E -s -v >> /etc/mdadm.conf将RAID配置保存下来，接着运行 update-initramfs -u（如果你的系统是Arch，则换成 mkinitcpio -p /etc/mkinitcpio.d/somekernel.preset）和 update-grub添加initrd/initramfs的hook

0b0010：格式化并挂载分区

此时的/dev/md0已经创建完成，但是还需要格式化并挂载才能使用

• 运行 mkfs.ext4 /dev/md0将其格式化成ext4格式：
  mke2fs 1.43.4 (31-Jan-2017) Creating filesystem with 20954624 4k blocks and 5242880 inodes Filesystem UUID: 6e458e01-b60f-4f64-8a48-c31d0fbd350e Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208, 4096000, 7962624, 11239424, 20480000 Allocating group tables: done Writing inode tables: done Creating journal (131072 blocks): done Writing superblocks and filesystem accounting information: done

  1 2 3 4 5 6 7 8 9 10 11

  mke2fs 1.43.4 (31-Jan-2017) Creating filesystem with 20954624 4k blocks and 5242880 inodes Filesystem UUID: 6e458e01-b60f-4f64-8a48-c31d0fbd350e Superblock backups stored on blocks:         32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,         4096000, 7962624, 11239424, 20480000   Allocating group tables: done Writing inode tables: done Creating journal (131072 blocks): done Writing superblocks and filesystem accounting information: done

• 接着 cp /etc/fstab /etc/fstab.bak备份一下挂载点文件
• 然后 echo '/dev/md0 /home ext4 barrier=0 0 0' >> /etc/fstab写入新分区信息，表示将其挂载到/home，当然你也可以将其挂载到其他地方
• 最后 mount -a重新挂载分区

0b0011：验证下效果

分别在不同的地方运行一下 fio -randrepeat=1 -ioengine=libaio -direct=1 -gtod_reduce=1 -name=test -filename=test -bs=4k -iodepth=64 -size=4G -readwrite=randrw -rwmixread=75测试4k读写

• 双高效云盘RAID0的阿里云（3862读1235写，由于是debian系统，输出格式可能和centos下的不太一样）：
  test: (g=0): rw=randrw, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=64 fio-2.16 Starting 1 process test: Laying out IO file(s) (1 file(s) / 4096MB) Jobs: 1 (f=1): [m(1)] [100.0% done] [15448KB/4940KB/0KB /s] [3862/1235/0 iops] [eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=10826: Thu Oct 4 02:58:47 2018 read : io=3070.4MB, bw=16957KB/s, iops=4239, runt=185406msec write: io=1025.8MB, bw=5664.1KB/s, iops=1416, runt=185406msec cpu : usr=1.26%, sys=6.08%, ctx=286721, majf=0, minf=9 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0% issued : total=r=785996/w=262580/d=0, short=r=0/w=0/d=0, drop=r=0/w=0/d=0 latency : target=0, window=0, percentile=100.00%, depth=64 Run status group 0 (all jobs): READ: io=3070.4MB, aggrb=16957KB/s, minb=16957KB/s, maxb=16957KB/s, mint=185406msec, maxt=185406msec WRITE: io=1025.8MB, aggrb=5664KB/s, minb=5664KB/s, maxb=5664KB/s, mint=185406msec, maxt=185406msec Disk stats (read/write): md0: ios=785569/262576, merge=0/0, ticks=0/0, in_queue=0, util=0.00%, aggrios=392998/131332, aggrmerge=0/36, aggrticks=5676188/234814, aggrin_queue=5911210, aggrutil=90.91% vdc: ios=393309/131035, merge=0/28, ticks=5435644/232016, in_queue=5667900, util=89.94% vdb: ios=392687/131630, merge=0/45, ticks=5916732/237612, in_queue=6154520, util=90.91%

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

  test: (g=0): rw=randrw, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=64 fio-2.16 Starting 1 process test: Laying out IO file(s) (1 file(s) / 4096MB) Jobs: 1 (f=1): [m(1)] [100.0% done] [15448KB/4940KB/0KB /s] [3862/1235/0 iops] [eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=10826: Thu Oct  4 02:58:47 2018   read : io=3070.4MB, bw=16957KB/s, iops=4239, runt=185406msec   write: io=1025.8MB, bw=5664.1KB/s, iops=1416, runt=185406msec   cpu          : usr=1.26%, sys=6.08%, ctx=286721, majf=0, minf=9   IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%      submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%      complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%      issued    : total=r=785996/w=262580/d=0, short=r=0/w=0/d=0, drop=r=0/w=0/d=0      latency   : target=0, window=0, percentile=100.00%, depth=64   Run status group 0 (all jobs):    READ: io=3070.4MB, aggrb=16957KB/s, minb=16957KB/s, maxb=16957KB/s, mint=185406msec, maxt=185406msec   WRITE: io=1025.8MB, aggrb=5664KB/s, minb=5664KB/s, maxb=5664KB/s, mint=185406msec, maxt=185406msec   Disk stats (read/write):     md0: ios=785569/262576, merge=0/0, ticks=0/0, in_queue=0, util=0.00%, aggrios=392998/131332, aggrmerge=0/36, aggrticks=5676188/234814, aggrin_queue=5911210, aggrutil=90.91%   vdc: ios=393309/131035, merge=0/28, ticks=5435644/232016, in_queue=5667900, util=89.94%   vdb: ios=392687/131630, merge=0/45, ticks=5916732/237612, in_queue=6154520, util=90.91%

• 默认的高效云盘阿里云（1971读646写，看起来也差不多是上面的一半）：
  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.1 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=7887KiB/s,w=2586KiB/s][r=1971,w=646 IOPS][eta 00m:00s] #阿里云在购买界面给出的20G大小的参考值是1960 IOPS，诚不我欺 test: (groupid=0, jobs=1): err= 0: pid=8814: Thu Oct 4 03:03:39 2018 read: IOPS=1965, BW=7863KiB/s (8052kB/s)(3070MiB/399800msec) bw ( KiB/s): min= 4488, max= 8682, per=100.00%, avg=7862.77, stdev=256.00, samples=799 iops : min= 1122, max= 2170, avg=1965.68, stdev=64.03, samples=799 write: IOPS=656, BW=2628KiB/s (2691kB/s)(1026MiB/399800msec) bw ( KiB/s): min= 1688, max= 3168, per=100.00%, avg=2627.91, stdev=174.95, samples=799 iops : min= 422, max= 792, avg=656.96, stdev=43.73, samples=799 cpu : usr=0.83%, sys=2.89%, ctx=365355, majf=0, minf=23 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0% issued rwt: total=785920,262656,0, short=0,0,0, dropped=0,0,0 latency : target=0, window=0, percentile=100.00%, depth=64 Run status group 0 (all jobs): READ: bw=7863KiB/s (8052kB/s), 7863KiB/s-7863KiB/s (8052kB/s-8052kB/s), io=3070MiB (3219MB), run=399800-399800msec WRITE: bw=2628KiB/s (2691kB/s), 2628KiB/s-2628KiB/s (2691kB/s-2691kB/s), io=1026MiB (1076MB), run=399800-399800msec Disk stats (read/write): vda: ios=787413/263981, merge=173/1680, ticks=24796074/718949, in_queue=25519686, util=99.93%

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.1 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=7887KiB/s,w=2586KiB/s][r=1971,w=646 IOPS][eta 00m:00s] #阿里云在购买界面给出的20G大小的参考值是1960 IOPS，诚不我欺 test: (groupid=0, jobs=1): err= 0: pid=8814: Thu Oct  4 03:03:39 2018    read: IOPS=1965, BW=7863KiB/s (8052kB/s)(3070MiB/399800msec)    bw (  KiB/s): min= 4488, max= 8682, per=100.00%, avg=7862.77, stdev=256.00, samples=799    iops        : min= 1122, max= 2170, avg=1965.68, stdev=64.03, samples=799   write: IOPS=656, BW=2628KiB/s (2691kB/s)(1026MiB/399800msec)    bw (  KiB/s): min= 1688, max= 3168, per=100.00%, avg=2627.91, stdev=174.95, samples=799    iops        : min=  422, max=  792, avg=656.96, stdev=43.73, samples=799   cpu          : usr=0.83%, sys=2.89%, ctx=365355, majf=0, minf=23   IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%      submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%      complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%      issued rwt: total=785920,262656,0, short=0,0,0, dropped=0,0,0      latency   : target=0, window=0, percentile=100.00%, depth=64   Run status group 0 (all jobs):    READ: bw=7863KiB/s (8052kB/s), 7863KiB/s-7863KiB/s (8052kB/s-8052kB/s), io=3070MiB (3219MB), run=399800-399800msec   WRITE: bw=2628KiB/s (2691kB/s), 2628KiB/s-2628KiB/s (2691kB/s-2691kB/s), io=1026MiB (1076MB), run=399800-399800msec   Disk stats (read/write):   vda: ios=787413/263981, merge=173/1680, ticks=24796074/718949, in_queue=25519686, util=99.93%

• 阿里云“企业入门级”SSD实例（3706读1255写）：
  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.1 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=14.5MiB/s,w=5020KiB/s][r=3706,w=1255 IOPS][eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=4218: Thu Oct 4 04:58:11 2018 read: IOPS=3601, BW=14.1MiB/s (14.8MB/s)(3070MiB/218228msec) bw ( KiB/s): min= 5808, max=17261, per=100.00%, avg=14436.27, stdev=920.02, samples=435 iops : min= 1452, max= 4315, avg=3609.03, stdev=230.00, samples=435 write: IOPS=1203, BW=4814KiB/s (4930kB/s)(1026MiB/218228msec) bw ( KiB/s): min= 1808, max= 5876, per=100.00%, avg=4824.62, stdev=344.82, samples=435 iops : min= 452, max= 1469, avg=1206.12, stdev=86.21, samples=435 cpu : usr=1.22%, sys=4.00%, ctx=160750, majf=0, minf=23 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0% issued rwt: total=785920,262656,0, short=0,0,0, dropped=0,0,0 latency : target=0, window=0, percentile=100.00%, depth=64 Run status group 0 (all jobs): READ: bw=14.1MiB/s (14.8MB/s), 14.1MiB/s-14.1MiB/s (14.8MB/s-14.8MB/s), io=3070MiB (3219MB), run=218228-218228msec WRITE: bw=4814KiB/s (4930kB/s), 4814KiB/s-4814KiB/s (4930kB/s-4930kB/s), io=1026MiB (1076MB), run=218228-218228msec Disk stats (read/write): vda: ios=785183/262515, merge=0/107, ticks=10244671/3660942, in_queue=13905813, util=99.99%

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.1 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=14.5MiB/s,w=5020KiB/s][r=3706,w=1255 IOPS][eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=4218: Thu Oct  4 04:58:11 2018    read: IOPS=3601, BW=14.1MiB/s (14.8MB/s)(3070MiB/218228msec)    bw (  KiB/s): min= 5808, max=17261, per=100.00%, avg=14436.27, stdev=920.02, samples=435    iops        : min= 1452, max= 4315, avg=3609.03, stdev=230.00, samples=435   write: IOPS=1203, BW=4814KiB/s (4930kB/s)(1026MiB/218228msec)    bw (  KiB/s): min= 1808, max= 5876, per=100.00%, avg=4824.62, stdev=344.82, samples=435    iops        : min=  452, max= 1469, avg=1206.12, stdev=86.21, samples=435   cpu          : usr=1.22%, sys=4.00%, ctx=160750, majf=0, minf=23   IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%      submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%      complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%      issued rwt: total=785920,262656,0, short=0,0,0, dropped=0,0,0      latency   : target=0, window=0, percentile=100.00%, depth=64   Run status group 0 (all jobs):    READ: bw=14.1MiB/s (14.8MB/s), 14.1MiB/s-14.1MiB/s (14.8MB/s-14.8MB/s), io=3070MiB (3219MB), run=218228-218228msec   WRITE: bw=4814KiB/s (4930kB/s), 4814KiB/s-4814KiB/s (4930kB/s-4930kB/s), io=1026MiB (1076MB), run=218228-218228msec   Disk stats (read/write):   vda: ios=785183/262515, merge=0/107, ticks=10244671/3660942, in_queue=13905813, util=99.99%

• 阿里云“本地SSD”实例，软RAID0（3W读1W写）：
  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.1 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=129MiB/s,w=42.4MiB/s][r=33.1k,w=10.9k IOPS][eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=29506: Tue Jan 29 10:09:21 2019 read: IOPS=34.0k, BW=133MiB/s (139MB/s)(3070MiB/23108msec) bw ( KiB/s): min=130288, max=137616, per=100.00%, avg=136045.74, stdev=1311.53, samples=46 iops : min=32572, max=34404, avg=34011.48, stdev=327.95, samples=46 write: IOPS=11.4k, BW=44.4MiB/s (46.6MB/s)(1026MiB/23108msec) bw ( KiB/s): min=43072, max=47280, per=100.00%, avg=45469.39, stdev=874.23, samples=46 iops : min=10768, max=11820, avg=11367.33, stdev=218.57, samples=46 cpu : usr=5.66%, sys=19.66%, ctx=7534, majf=0, minf=8 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0% issued rwt: total=785920,262656,0, short=0,0,0, dropped=0,0,0 latency : target=0, window=0, percentile=100.00%, depth=64 Run status group 0 (all jobs): READ: bw=133MiB/s (139MB/s), 133MiB/s-133MiB/s (139MB/s-139MB/s), io=3070MiB (3219MB), run=23108-23108msec WRITE: bw=44.4MiB/s (46.6MB/s), 44.4MiB/s-44.4MiB/s (46.6MB/s-46.6MB/s), io=1026MiB (1076MB), run=23108-23108msec Disk stats (read/write): md0: ios=779768/265309, merge=0/0, ticks=0/0, in_queue=0, util=0.00%, aggrios=392960/133444, aggrmerge=0/271, aggrticks=622454/1578, aggrin_queue=589912, aggrutil=81.70% vdc: ios=392952/133448, merge=0/0, ticks=575400/1612, in_queue=548200, util=75.15% vdb: ios=392968/133440, merge=0/543, ticks=669508/1544, in_queue=631624, util=81.70%

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.1 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=129MiB/s,w=42.4MiB/s][r=33.1k,w=10.9k IOPS][eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=29506: Tue Jan 29 10:09:21 2019    read: IOPS=34.0k, BW=133MiB/s (139MB/s)(3070MiB/23108msec)    bw (  KiB/s): min=130288, max=137616, per=100.00%, avg=136045.74, stdev=1311.53, samples=46    iops        : min=32572, max=34404, avg=34011.48, stdev=327.95, samples=46   write: IOPS=11.4k, BW=44.4MiB/s (46.6MB/s)(1026MiB/23108msec)    bw (  KiB/s): min=43072, max=47280, per=100.00%, avg=45469.39, stdev=874.23, samples=46    iops        : min=10768, max=11820, avg=11367.33, stdev=218.57, samples=46   cpu          : usr=5.66%, sys=19.66%, ctx=7534, majf=0, minf=8   IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%      submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%      complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%      issued rwt: total=785920,262656,0, short=0,0,0, dropped=0,0,0      latency   : target=0, window=0, percentile=100.00%, depth=64   Run status group 0 (all jobs):    READ: bw=133MiB/s (139MB/s), 133MiB/s-133MiB/s (139MB/s-139MB/s), io=3070MiB (3219MB), run=23108-23108msec   WRITE: bw=44.4MiB/s (46.6MB/s), 44.4MiB/s-44.4MiB/s (46.6MB/s-46.6MB/s), io=1026MiB (1076MB), run=23108-23108msec   Disk stats (read/write):     md0: ios=779768/265309, merge=0/0, ticks=0/0, in_queue=0, util=0.00%, aggrios=392960/133444, aggrmerge=0/271, aggrticks=622454/1578, aggrin_queue=589912, aggrutil=81.70%   vdc: ios=392952/133448, merge=0/0, ticks=575400/1612, in_queue=548200, util=75.15%   vdb: ios=392968/133440, merge=0/543, ticks=669508/1544, in_queue=631624, util=81.70%

• 我自己本地的测试机（12900读4247写）：
  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.3 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=50.5MiB/s,w=16.6MiB/s][r=12.9k,w=4247 IOPS][eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=10971: Thu Oct 4 04:41:27 2018 read: IOPS=12.3k, BW=48.2MiB/s (50.6MB/s)(3070MiB/63671msec) bw ( KiB/s): min=28248, max=57680, per=99.94%, avg=49341.31, stdev=4138.57, samples=127 iops : min= 7062, max=14420, avg=12335.32, stdev=1034.64, samples=127 write: IOPS=4125, BW=16.1MiB/s (16.9MB/s)(1026MiB/63671msec) bw ( KiB/s): min= 9635, max=19128, per=99.95%, avg=16491.75, stdev=1361.83, samples=127 iops : min= 2408, max= 4782, avg=4122.91, stdev=340.48, samples=127 cpu : usr=6.75%, sys=24.72%, ctx=394816, majf=0, minf=135 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0% issued rwts: total=785920,262656,0,0 short=0,0,0,0 dropped=0,0,0,0 latency : target=0, window=0, percentile=100.00%, depth=64 Run status group 0 (all jobs): READ: bw=48.2MiB/s (50.6MB/s), 48.2MiB/s-48.2MiB/s (50.6MB/s-50.6MB/s), io=3070MiB (3219MB), run=63671-63671msec WRITE: bw=16.1MiB/s (16.9MB/s), 16.1MiB/s-16.1MiB/s (16.9MB/s-16.9MB/s), io=1026MiB (1076MB), run=63671-63671msec Disk stats (read/write): dm-0: ios=787183/262248, merge=0/0, ticks=3056810/970151, in_queue=4028413, util=99.97%, aggrios=785350/262402, aggrmerge=3793/484, aggrticks=3052880/970600, aggrin_queue=4023576, aggrutil=99.94% sda: ios=785350/262402, merge=3793/484, ticks=3052880/970600, in_queue=4023576, util=99.94%

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

  test: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64 fio-3.3 Starting 1 process test: Laying out IO file (1 file / 4096MiB) Jobs: 1 (f=1): [m(1)][100.0%][r=50.5MiB/s,w=16.6MiB/s][r=12.9k,w=4247 IOPS][eta 00m:00s] test: (groupid=0, jobs=1): err= 0: pid=10971: Thu Oct  4 04:41:27 2018    read: IOPS=12.3k, BW=48.2MiB/s (50.6MB/s)(3070MiB/63671msec)    bw (  KiB/s): min=28248, max=57680, per=99.94%, avg=49341.31, stdev=4138.57, samples=127    iops        : min= 7062, max=14420, avg=12335.32, stdev=1034.64, samples=127   write: IOPS=4125, BW=16.1MiB/s (16.9MB/s)(1026MiB/63671msec)    bw (  KiB/s): min= 9635, max=19128, per=99.95%, avg=16491.75, stdev=1361.83, samples=127    iops        : min= 2408, max= 4782, avg=4122.91, stdev=340.48, samples=127   cpu          : usr=6.75%, sys=24.72%, ctx=394816, majf=0, minf=135   IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%      submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%      complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.1%, >=64=0.0%      issued rwts: total=785920,262656,0,0 short=0,0,0,0 dropped=0,0,0,0      latency   : target=0, window=0, percentile=100.00%, depth=64   Run status group 0 (all jobs):    READ: bw=48.2MiB/s (50.6MB/s), 48.2MiB/s-48.2MiB/s (50.6MB/s-50.6MB/s), io=3070MiB (3219MB), run=63671-63671msec   WRITE: bw=16.1MiB/s (16.9MB/s), 16.1MiB/s-16.1MiB/s (16.9MB/s-16.9MB/s), io=1026MiB (1076MB), run=63671-63671msec   Disk stats (read/write):     dm-0: ios=787183/262248, merge=0/0, ticks=3056810/970151, in_queue=4028413, util=99.97%, aggrios=785350/262402, aggrmerge=3793/484, aggrticks=3052880/970600, aggrin_queue=4023576, aggrutil=99.94%   sda: ios=785350/262402, merge=3793/484, ticks=3052880/970600, in_queue=4023576, util=99.94%

• 至于Amazon AWS、Vultr跟DO不知为何跑不起来；Sakura的入门机型则是6046读1997写。（我开始编辑这篇文章的时候，以大坊开始同步，写道这里的时候同步已经完成，偶也！）

0b0100：结论

• 至少在阿里云上面，2个盘的RAID0的确可以有效加快速度，相比SSD云盘，用2个一半容量的“高效云盘”做软RAID的确可以达到类似性能，Instagram也干过类似的事
• 从价格上来说，阿里云的SSD云盘差不多比“高效云盘”贵了一倍左右，如果手上的确比较紧张，可以考虑这个方案
• RAID0本身是非常不安全的，任何一个节点出问题整个阵列就挂，不过根据阿里云官方的说法，云盘“高性能、持久性、高可靠”，所以我认为可以先观察观察