RAID 0, RAID 1, RAID 5, RAID 10 Explained with Diagrams

by Ramesh Natarajan on August 10, 2010

RAID stands for Redundant Array of Inexpensive (Independent) Disks.

On most situations you will be using one of the following four levels of RAIDs.

  • RAID 0
  • RAID 1
  • RAID 5
  • RAID 10 (also known as RAID 1+0)

This article explains the main difference between these raid levels along with an easy to understand diagram.

In all the diagrams mentioned below:

  • A, B, C, D, E and F – represents blocks
  • p1, p2, and p3 – represents parity


Following are the key points to remember for RAID level 0.

  • Minimum 2 disks.
  • Excellent performance ( as blocks are striped ).
  • No redundancy ( no mirror, no parity ).
  • Don’t use this for any critical system.


Following are the key points to remember for RAID level 1.

  • Minimum 2 disks.
  • Good performance ( no striping. no parity ).
  • Excellent redundancy ( as blocks are mirrored ).


Following are the key points to remember for RAID level 5.

  • Minimum 3 disks.
  • Good performance ( as blocks are striped ).
  • Good redundancy ( distributed parity ).
  • Best cost effective option providing both performance and redundancy. Use this for DB that is heavily read oriented. Write operations will be slow.


Following are the key points to remember for RAID level 10.

  • Minimum 4 disks.
  • This is also called as “stripe of mirrors”
  • Excellent redundancy ( as blocks are mirrored )
  • Excellent performance ( as blocks are striped )
  • If you can afford the dollar, this is the BEST option for any mission critical applications (especially databases).

Additional RAID Tutorials:

If you enjoyed this article, you might also like..

  1. 50 Linux Sysadmin Tutorials
  2. 50 Most Frequently Used Linux Commands (With Examples)
  3. Top 25 Best Linux Performance Monitoring and Debugging Tools
  4. Mommy, I found it! – 15 Practical Linux Find Command Examples
  5. Linux 101 Hacks 2nd Edition eBook Linux 101 Hacks Book

Bash 101 Hacks Book Sed and Awk 101 Hacks Book Nagios Core 3 Book Vim 101 Hacks Book

{ 228 comments… read them below or add one }

201 Justin Goldberg March 10, 2014 at 9:21 am

I think the only reason to use Raid 0 is a temp or scratch drive, such as photoshop scratch drive, ram drive, browser cache, etc…

202 Josh March 14, 2014 at 10:52 am

It is incorrect to state Raid 10 is the same as Raid 1+0 which is a confusion with Raid 0+1.

Raid 10 is always referred to as raid 10 never as 1+0.
Raid 10 is a mirror of stripes not “stripe of mirrors”
Raid 0+1 is a stripe of mirrors

You are confused between Raid 10 and Raid 0+1

Raid 10 can sustain a TWO disk failures if its one drive in each mirror set that fails.
Raid 0+1 with the loss of a single drive reverts to a Raid0 array.

Please correct your misinformation. Thanks

203 Gerald March 14, 2014 at 6:30 pm

Josh, you are right that RAID0+1 & RAID1+0 have become confused as different manufacturers used the two forms to mean different things. So the terms became ambiguous, and there is a move to deprecate these terms. AS you rightly point out Striped-Mirrors have redundancy, whereas Mirrored-Stripes do not.

That was why the term RAID10 was coined and in general everybody has adopted this as meaning Striped Mirrors, ie at the bottom level are Mirrorsets that have been Striped, although last time i looked Dell had some documents that defined it as mirrored-stripes, which is clearly incorrect.

So Your definition of RAID10 is incorrect, the order of the digits is the clue, so RAID10 is RAID-1 RAIDsets that are then Striped (RAID-0), this definition is also followed by RAID50 (RAID-5 + RAID-0) and RAID60 (RAID-6 + RAID-0)

You are also incorrect in stating that RAID10 can sustain TWO disk failures, in fact a Striped-Mirrorset can sustain multiple disk failures depending on how many Mirrorsets the stripe is across and crucially that two failures are not in the same Mirrorset

204 samiuddin March 16, 2014 at 12:47 am

Excellent Explanation! Keep up the Good Work

205 Josh March 17, 2014 at 2:33 pm

” You are also incorrect in stating that RAID10 can sustain TWO disk failures”

When we talk about raid we talk about the minimum requirements. For example 2 disks for raid1, three disks for raid5 and 4 disks for raid 0+1 and raid10

My statement is entirely accurate if you consider a 4 disk system as I described and as you list above in your raid 10 example.

” Your definition of RAID10 is incorrect, the order of the digits is the clue, so RAID10 is RAID-1 RAIDsets that are then Striped (RAID-0),”

You are absolutely right. That’s what I get for making comments in a hurry :-)

The best way to think about both raid10 and raid0+1 is you take the last digit, in the case of raid 10 that is stripping and apply it to the first digit mirroring which makes it a stripe of a set of mirrors. Conversely you have with raid 0+1 you have a mirror of striped disks.

And thank you for the opportunity to play the babbling tech :-)

206 Gerald March 19, 2014 at 11:33 pm

Josh, just forget about RAID-0+1 (and 1+0 as well) this notation is now not in current use. Just use RAID10, most people know that this means mirrors that are striped.

I was always taught (back in the 90′s) that 0+1 was striped mirrors, but people got it into their head that you read the digits left to right and that meant it was stripes that were mirrored, that’s why RAID10 came along to remove the ambiguity

Ps note only one “p” in stripe, stripping is something else. :-)

207 theo April 4, 2014 at 2:43 pm

Hmmmm . . . all I know is this, if you are “stripping” it will no doubt be followed by a “Raid”. :)

208 Gerald April 9, 2014 at 2:12 pm


209 perrie iles April 30, 2014 at 7:00 am

Dear sir,
i am moving to an Asrock Z77 extreme 11, it supports 8 drives in raid 0,1,1E or 10 – i have 8 x 4tb (plus others to fill the other 6 sata ports) – what schemas would be optimum to increase data security without losing significant space or reducing write times to far – ( i currently run 12 drives +2 ssd +1 odd )

210 Charan May 14, 2014 at 8:12 am

What is Disk1 and Disk2 fails in RAID10?. Data will be incomplete? Am I right?

211 Gerald May 28, 2014 at 1:06 pm

Correct, if both disks in one of the Mirrorsets fail simultaneously, then the Data will be incomplete.

Incidentally a RAID10 can sustain n Disk failures where n is the number of Mirrorsets in the stripe.

212 Gerald May 28, 2014 at 1:12 pm

RAID10 would give you the optimum data security and the best performance but at with only a 50% space utilisation.

RAID-6 would give you excellent data security, with excellent Utilisation, but with the potential for the worst performance, especially if you usage is write biased.

213 methun June 10, 2014 at 6:27 am

which is better RAID 5, RAID 10,RAID 50? and why?

214 Gerald June 11, 2014 at 12:36 am

Its not a case of which is better its a case of which is more suitable!

RAID-5 – Gives good usable space, good read speed, but is not good for a write biased workload. Not recommended for use with very large disks due to rebuild times (use RAID10 or RAID-6 instead)

RAID50 – ditto, read even faster, still has write issue.

RAID10 – Best performance for Read and Write, but lowest useable space (50%)

NB useable space is “Total disk space” – “redundancy overhead”

215 Nation Ali June 11, 2014 at 5:26 am

We are using 5 disk in RAID. unfortunately we remove all drive, Now we re-install these drive but O.S not booting. how we are re-configure these disk without losing Data.

We are using HP ML 350 SAS also have disk HP software DVD.

216 rathish July 2, 2014 at 12:54 am

Hi to all, I have a doubt in linux how to know which type of raid is configured Is there any cmd.

217 Gerald July 5, 2014 at 3:27 pm

@Nation Ali – Normally with HP RAID controllers you can take the disks out, juggle them into any random order and replace and they will work! However your question is not suitable for this forum please use something like Experts-Exchange (

@rathish – this will depend on which Linux distro you are using! However your question is not suitable for this forum please use something like Experts-Exchange (

218 Dino Davis August 18, 2014 at 10:09 am

Hello, I am a newbie to the server administration and I need to know more cloud storage and RAID Controllers. In which curriculum where I can learn more about this?

219 rainbow September 10, 2014 at 3:28 am

Can I do this with SCSI Drives too? It is an older system.How i can check if it is possible?

Thank you in advanced!

220 chandan September 12, 2014 at 6:19 am

Ya it’s good help to fresher…

221 Salih September 25, 2014 at 12:57 am

Dear Ramesh,

Thanks for giving clarity to the RAID Levels with diagrams.

222 Gerald September 30, 2014 at 1:59 am

@Rainbow – Of course! Depends on hardware. If you have a RAID controller, find a manaula for it, if not use Windows/Linux software RAID.
Avoid RAID-5 or -6 via software though (big processing overhead)

and DON’T FORGET to BACKUP EVERYTHING before experimenting.

223 p sai krishna balaji October 5, 2014 at 11:31 pm

hi this is balaji your linux tips are very intilsent brain ok sir


224 Keshav October 29, 2014 at 7:15 am


I have setup raid1 on AWS EC2, and on testing found that its working fine.

Concern is that I am doing this first time, I dont know how to host my site to work with raid1?

Thanks in advance.

225 Doug December 13, 2014 at 10:27 am

You say with Raid-5 that writes are slow. Is that relative to other configurations with the same number of drives or in general? Specifically, how does a 3 disk Raid-5 compare to a 2 disk Raid-1 system? I was told that a 3 disk Raid-5 system will perform both reads and writes better as 2 blocks can be read/written in one cycle

In general, does this logic apply equally to SSD drives?


226 abhisheklohar January 2, 2015 at 11:41 am

RAID USED for only server client and provide the gud performance

227 Gerald January 4, 2015 at 11:01 am

@Keshav – What is it you are trying to do?

@abhisheklohar – Is this a Comment or a Question

228 Gerald January 14, 2015 at 1:22 pm

To do an update transaction, you need to Read, Update & then Write

Single Read to RAID Controller and if it is clever enough it will issue the read to both disks and the return the data from the first one to deliver (ie reduced Latency).
Update the data and a single Write to the controller which then writes to both disks in parallel.
So, Max 3 IOPS (simple example case)

RAID-5 (3 Disk RAIDset)
Single Read to RAID controller, and probably a single Read from disk.
Update data and then a single Write to the RAID Controller, but the Controller has to read all the other disks (including the Parity) Ie 2 * IOPS, XOR all the disk data with the Update and write the Update and the Parity ie 2 * IOPS
So, Min 5 IOPS to do the same job.

Leave a Comment

Previous post:

Next post: