============ OS and Kernel
Linux server4.localdomain 4.15.7 #1 SMP Sun Mar 4 16:43:40 CET 2018 x86_64 x86_64 x86_64 GNU/Linux
============ filesystem used
Filesystem     Type      Size  Used Avail Use% Mounted on
devtmpfs       devtmpfs   30G     0   30G   0% /dev
tmpfs          tmpfs      30G     0   30G   0% /dev/shm
tmpfs          tmpfs      30G  9.2M   30G   1% /run
tmpfs          tmpfs      30G     0   30G   0% /sys/fs/cgroup
/dev/sda3      ext3       50G   44G  3.3G  93% /
/dev/md0       ext4      5.5T   89M  5.2T   1% /run/media/forstner/RAID10
/dev/sda1      ext3      976M  321M  605M  35% /boot
/dev/sda2      ext3      161G  122M  153G   1% /home
tmpfs          tmpfs     5.9G   28K  5.9G   1% /run/user/1000

iozone -version
‚Iozone‘ Filesystem Benchmark Program

Version $Revision: 3.424 $
Compiled for 64 bit mode.

Original Author: William Norcott (wnorcott@us.oracle.com)
4 Dunlap Drive
Nashua, NH 03060

Enhancements: Don Capps (capps@iozone.org)
7417 Crenshaw
Plano, TX 75025

Copyright 1991, 1992, 1994, 1998, 1999, 2002 William D. Norcott

License to freely use and distribute this software is hereby granted
by the author, subject to the condition that this copyright notice
remains intact. The author retains the exclusive right to publish
derivative works based on this work, including, but not limited to,
revised versions of this work

Other contributors:

Don Capps (Network Appliance) capps@iozone.org

benchmark results

ext4 is fast, but xfs is even faster – but remember: have good backup strategy in place if you NEED the SPEED, because it comes with (IMHO) with the disadvantage of not being able to undelete accidentally deleted files.

Expect shit to happen.

ssd+ext3

mkfs.ext3 -V
mke2fs 1.42.9 (28-Dec-2013)
Using EXT2FS Library version 1.42.9

SATA attached Model=Samsung SSD 850 PRO 256GB, FwRev=EXM04B6Q, SerialNo=S39KNX0J917798M

/dev/sda (ext3)

time iozone -a; # run harddisk benchmark

iozone test complete.

real 4m34.297s
user 0m3.369s
sys 1m7.935s

raid10:

raid10 is multiple raid1 (two or more redundant mirrored disks) combined in a raid0. hence raid1+0 = raid10, it combines speed with redundancy.

cat /proc/mdstat 
Personalities : [raid10] 
md0 : active raid10 sdb1[0] sdc1[1] sdd1[2] sdf1[4] sde1[3] sdh1[6](S) sdg1[5]
      5860147200 blocks super 1.2 512K chunks 2 near-copies [6/6] [UUUUUU]
      bitmap: 2/44 pages [8KB], 65536KB chunk

it is read from/written to 3x sets of RAID1 composed of two 2TB Western Digital Disks at the same time, so theoretically you should get 3x times the speed of a WD2005FBYZ.

Attached are those disks to a LSI MegaRaid 1068E onboard controller that is jumpered to SATA mode (the controller itself is not used for RAID)

H/W path Device Class Description
=============================================================
1. /0/100/8/0/0.0.0 /dev/sdb disk 2TB WDC WD2005FBYZ-0
2. /0/100/8/0/0.1.0 /dev/sdc disk 2TB WDC WD2004FBYZ-0
3. /0/100/8/0/0.2.0 /dev/sdd disk 2TB WDC WD2005FBYZ-0
4. /0/100/8/0/0.3.0 /dev/sde disk 2TB WDC WD2002FYPS-0
5. /0/100/8/0/0.4.0 /dev/sdf disk 2TB WDC WD2002FYPS-0
6. /0/100/8/0/0.5.0 /dev/sdg disk 2TB WDC WD2003FYPS-2
7. /0/100/8/0/0.6.0 /dev/sdh disk 2TB WDC WD2003FYYS-0 (spare)

raid10+ext3

md0 (ext3)

mkfs.ext3 -V
mke2fs 1.42.9 (28-Dec-2013)
Using EXT2FS Library version 1.42.9

real 5m2.363s
user 0m3.552s
sys 1m12.694s

second run:

real 4m45.454s
user 0m3.555s
sys 1m7.884s

raid10+ext4

md0 (ext4)

mkfs.ext4 -V
mke2fs 1.42.9 (28-Dec-2013)
Using EXT2FS Library version 1.42.9

 

time iozone -a; # run harddisk benchmark

iozone test complete

real 4m33.547s
user 0m3.683s
sys 1m5.642s

second run:

real 4m43.598s
user 0m3.491s
sys 1m4.842s

raid10+xfs

mkfs.xfs -V
mkfs.xfs version 4.5.0

mkfs.xfs -f /dev/md0 -L „RAID10“; # create xfs filesystem

time iozone -a; # run harddisk benchmark

iozone test complete.

real 4m19.871s
user 0m3.520s
sys 1m3.004s

second run:

real 4m20.964s
user 0m4.002s
sys 1m2.615s

 

admin