Configuring a Software RAID on a Linux General Purpose Cloud Server


This article will demonstrate how to put multiple data disks on a General Purpose server into a RAID level 0.  We will mount a single data disk and perform read and write tests, then continue by putting two data disks into a RAID 0.

A RAID 0 stripes (combines) two disks to look like one drive to the system, usually increasing performance (particularly read access).  While a RAID level of 0 only stripes the disks and offers no data redunancy, the hypervisor (host server) is backed by a RAID 10.  This RAID 10 provides redunancy for your data on the backend.

Prerequisites

Creating a RAID level 0 requires a General Purpose server with at least two data disks.

To configure the RAID we will use the mdadm software RAID utility, which may need to be installed.

Redhat/Centos/Fedora/Scientific Linux mdadm installation

sudo yum install mdadm

Ubuntu/Debian mdadm installation

sudo apt-get update
sudo apt-get install mdadm

Identifying available data disks

Disks on Linux are referenced using their device name.  The/dev/xvda device is your system disk and contain your operating system.  We will use the fdisk utility to identify your data disk devices (usually /dev/xvde and /dev/xvdf).

sudo fdisk -l

The output will list full details about the disks attached to the system.  One disk entry would look like this:

Disk /dev/xvde: 322.1 GB, 322122547200 bytes
255 heads, 63 sectors/track, 39162 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x000ea711

    Device Boot      Start         End      Blocks   Id  System
    /dev/xvde1             1       37349   300000000   83  Linux

The first line shows the device name and size.  The last line of the block shows any partitions currently configured on the drive, with their device names (like /dev/xvde1).

Creating the RAID 0

Now that we have the device names of our data disks we can start to provision the RAID 0 with the following commands.

sudo mdadm --create /dev/md0 --level=0 --raid-devices=2 /dev/xvde /dev/xvdf

You will get a warning if any partitions exist on the disks being provisioned.  Confirm that the partitions can be overwritten and mdadm will create the RAID.

Verify the RAID 0 by checking the /proc/mdstat system file.

cat /proc/mdstat

The file's contents should look something like this:

Personalities : [raid0]
    md0 : active raid0 xvdf[1] xvde[0]
          629144576 blocks super 1.2 512k chunks

     unused devices: <none>

The beginning of the second line in the example shows the device name (md0), which means we can reference the array as /dev/md0.

Partitioning the RAID 0

Once the array is created we can partition it to make a file system.  We'll use the fdisk utility to create a single partition on the RAID 0.

sudo fdisk /dev/md0

After displaying some information about the device we're editing (the array), you'll see a command prompt.

We'll need to create a new partition using all available disk space, then write the changes to the array's partition table.

Start by entering "n" for new partition, then "p" for a primary partition, and "1" for the partition number.  Just hit enter when asked for the starting and ending cylinders so fdisk will use the defaults, filling the disk with the single partition.  Once the partition is created enter "w" to write the changes to the disk.

The process will look something like this:

Command (m for help): n

Command action
   e   extended
   p   primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-157286144, default 257): 
Using default value 257
Last cylinder, +cylinders or +size{K,M,G} (257-157286144, default 157286144): 
Using default value 157286144

Command (m for help): w
The partition table has been altered!

Calling ioctl() to re-read partition table.
Syncing disks.

After writing the partition table, fdisk should exit.  Running fdisk again should let you see the array and its new partition (along with the partition's device name).

sudo fdisk -l

Creating the file system on the RAID 0

For the purposes of this article we'll use ext4 for our demonstration and performance testing.

Note:  If you use ext4 on your array, make sure your system supports it.  Recent distributions support ext4 by default, but if you're using an older base operating system (like CentOS 5.3) the included kernel and disk formatting utility may not support ext4.  In this case, it's safer to use ext3 (with the mkfs.ext3 command).

Creating the file system will be easy and seamless.  Run the appropriate mkfs command for the file system (mkfs.ext4 to format as ext4, for example) on the array's partition, usually /dev/md0p1.

sudo mkfs.ext4 /dev/md0p1

With the new file system created you are free to mount the array to any mount point that you would like.  Edit your /etc/fstab file to add a line for the new disk.  If we wanted to mount the disk on our example system on /var/lib/mysql, we would add the following line:

/dev/md0p1          /var/lib/mysql  ext4    defaults,noatime      0      2

To mount the RAID 0 after saving the fstab file, run:

sudo mount -a

Performance testing

For the performancing testing on our example RAID 0 we ran several read and write tests, with the following results (including commands and output):

RAID level 0 with data disk 60 GB performance server 

[READ] /dev/md0

[root@performance-60GB ~]# echo 3 > /proc/sys/vm/drop_caches 

[root@performance-60GB ~]# dd if=/mnt/speed.file of=/dev/null bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.166875 s, 307 MB/s

[root@performance-60GB ~]# echo 3 > /proc/sys/vm/drop_caches 

[root@performance-60GB ~]# dd if=/mnt/speed.file of=/dev/null bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.16641 s, 308 MB/s

[root@performance-60GB ~]# echo 3 > /proc/sys/vm/drop_caches 

[root@performance-60GB ~]# dd if=/mnt/speed.file of=/dev/null bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.166675 s, 307 MB/s

[Write] /dev/md0

[root@performance-60GB ~]# dd if=/dev/zero of=/mnt/speed.file bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.343796 s, 149 MB/s

[root@performance-60GB ~]# rm -fv /mnt/speed.file 
removed `/mnt/speed.file'

[root@performance-60GB ~]# dd if=/dev/zero of=/mnt/speed.file bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.343648 s, 149 MB/s

[root@performance-60GB ~]# rm -fv /mnt/speed.file 
removed `/mnt/speed.file'
[root@performance-60GB ~]# dd if=/dev/zero of=/mnt/speed.file bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.345652 s, 148 MB/s

[root@performance-60GB ~]# cat /proc/mdstat
Personalities : [raid0] 
md0 : active raid0 xvde[1] xvdf[0]
      629144576 blocks super 1.2 512k chunks
      
unused devices: <none>


No RAID level 0 


[READ] 

[root@performance-60GB ~]# echo 3 > /proc/sys/vm/drop_caches 

[root@performance-60GB ~]# dd if=/mnt/speed.file of=/dev/null bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.195058 s, 262 MB/s
[root@performance-60GB ~]# echo 3 > /proc/sys/vm/drop_caches 

[root@performance-60GB ~]# dd if=/mnt/speed.file of=/dev/null bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.198602 s, 258 MB/s

[root@performance-60GB ~]# echo 3 > /proc/sys/vm/drop_caches 

[root@performance-60GB ~]# dd if=/mnt/speed.file of=/dev/null bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.199001 s, 257 MB/s


[WRITE]

[root@performance-60GB ~]# dd if=/dev/zero of=/mnt/speed.file bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.337723 s, 152 MB/s

[root@performance-60GB ~]# rm -fv /mnt/speed.file 
removed `/mnt/speed.file'

[root@performance-60GB ~]# dd if=/dev/zero of=/mnt/speed.file bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.34109 s, 150 MB/s

[root@performance-60GB ~]# rm -fv /mnt/speed.file 
removed `/mnt/speed.file'

[root@performance-60GB ~]# dd if=/dev/zero of=/mnt/speed.file bs=1024 count=50000
50000+0 records in
50000+0 records out
51200000 bytes (51 MB) copied, 0.33958 s, 151 MB/s

Data Results :

RAID 0 -

Read Avg : 307 MB/s
Write Avg : 148 MB/s

No RAID -

Read Avg : 259 MB/s
Write Avg : 151 MB/s

Our results from the testing showed a 16% increase in reads while utilizing (2) SSD's in a RAID 0, with a 2% decrease in write performance. The performance gains from the reads is substantial enough to warrant utilizing the RAID 0 for most purposes, but if you're running an application that performs more writes than reads you may benefit more from using the data disks stand-alone instead of going with the RAID 0 option.



Was this content helpful?




© 2014 Rackspace US, Inc.

Except where otherwise noted, content on this site is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License


See license specifics and DISCLAIMER