The importance of back ups

Sorry I am being so dense on this subject. I feel I dont really even know what RAID is to be lost!


RAID devices contain a number of physical disks. The data is shared across the disks in various magic ways so that if a physical disk fails you don't lose any data (EXCEPT FOR RAID 0!!!). Because of the way they work, many people believe that a RAID contains several copies of a file (it kind of does sometimes but not really) and think "oo, my RAID has 2 (or more) copies of the file so one of them backs up the other". It doesn't.

Just think of a RAID as a regular disk drive that is less likely to fail than a normal one (unless it's RAID 0....). So each RAID has one copy of the file (even though that may not be literally true).

[BTW the number after the RAID tells you in a typically techie way how the multiple drives are configured. 0 means they are configured for danger, 1 means basic safety, 5 means more expensive safety and 10 means you're really rich. I forget what 6 means.....]
 
right thanks. So if you have a drive on your computer and then you have a raid(6) drive sat next to your computer and are using it as a backup of the physical drive of your computer then thats good.

In fact better than having a non raid drive as the secondary drive for the backups due to it being safer.

Othe than the fact it's still not a good idea for that to be the limitation of your setup because fire theft are still possible - but as the first stage in your backup using the raid drive is better than not using a raid drive

Is that correct?
 
right thanks. So if you have a drive on your computer and then you have a raid(6) drive sat next to your computer and are using it as a backup of the physical drive of your computer then thats good.

In fact better than having a non raid drive as the secondary drive for the backups due to it being safer.

Othe than the fact it's still not a good idea for that to be the limitation of your setup because fire theft are still possible - but as the first stage in your backup using the raid drive is better than not using a raid drive

Is that correct?

Sure. Although there are cost and speed implications ;)

http://en.wikipedia.org/wiki/Standard_RAID_levels
 
The clue is in the acronym. RAID = Redundant Array of Inexpensive Disks. It is there for redundancy, not backup. I.e. if a disk fails, you can normally continue working.

RAID 0 is for speed, and halves/thirds/quarters (however many disks you have in the array) the mean time between failures. It is misnamed as there is no redundancy.
RAID 1 is the simple but expensive option (I'll keep a copy of everything)
RAID 5 is the thinking mans RAID - N disks, with 1 spare to replace any that fails. This is what most sensible people use at home.
RAID 6 is the rich thinking mans RAID - N disks with 2 spare to replace any 2 that die in one go.
RAID 10 is a combination of RAID 0 and RAID 1. For specialists only who require the ultimate in uptime and speed (think datacentres) as it is expensive.

Think of a RAID "array" as a single disk as far as you are concerned (even if it has multiple disks in it) and you will be fine.
 
Sure. Although there are cost and speed implications ;)

http://en.wikipedia.org/wiki/Standard_RAID_levels

thanks!

So when it is used in the way I suggested above it is actually a backup.

but yet people are still saying it's not a backup. I think I now understand why.

i read this article

http://www.petemarovichimages.com/2013/11/24/never-use-a-raid-as-your-backup-system/

the bit that made the penny drop was this

"A BACKUP needs to be a complete and recoverable copy of your data that resides on a separate hard drive possibly even a RAID"

when folk in this theead are saying RAID is not a backup what they mean is having just a RAID drive with everything on isn't a backup system - but using a RAID drive as a seperate bakcup drive is fine. I think Neil already said that above but thats back when I was confused.

I'm pretty sure I get it now. Whenever someone says "RAID is not a backup" they mean not as your primary drive. They just dont say that
 
but yet people are still saying it's not a backup. I think I now understand why.
used as the ONLY copy RAID is not a backup. some people assume that because raid offers some form of protection against disk failure that they are protected against other factors. when in fact this is wrong.

if the RAID is your 2nd/3rd/4th..etc copy then this is fine.

the person i quoted originally has not stated whether it was their only copy, hence my comment about hoping that was not the case :)
 
used as the ONLY copy RAID is not a backup. some people assume that because raid offers some form of protection against disk failure that they are protected against other factors. when in fact this is wrong.

if the RAID is your 2nd/3rd/4th..etc copy then this is fine.

the person i quoted originally has not stated whether it was their only copy, hence my comment about hoping that was not the case :)

yes! ok i get it. thanks for being patient

"used as the ONLY copy RAID is not a backup" I move to ensure everyone states this when they say RAID isn't a backup. saves the confusion :-)
 
right thanks. So if you have a drive on your computer and then you have a raid(6) drive sat next to your computer and are using it as a backup of the physical drive of your computer then thats good.

In fact better than having a non raid drive as the secondary drive for the backups due to it being safer.

Othe than the fact it's still not a good idea for that to be the limitation of your setup because fire theft are still possible - but as the first stage in your backup using the raid drive is better than not using a raid drive

Is that correct?

Correct, but for important information you need an off-site backup. I personally use Amazon Glacier as it's relatively cheap and the only downside I can see is the 8 hours it takes to retrieve data - it being stored on tape and needing to be reloaded.
 
The clue is in the acronym. RAID = Redundant Array of Inexpensive Disks. It is there for redundancy, not backup. I.e. if a disk fails, you can normally continue working.

RAID 0 is for speed, and halves/thirds/quarters (however many disks you have in the array) the mean time between failures. It is misnamed as there is no redundancy.
RAID 1 is the simple but expensive option (I'll keep a copy of everything)
RAID 5 is the thinking mans RAID - N disks, with 1 spare to replace any that fails. This is what most sensible people use at home.
RAID 6 is the rich thinking mans RAID - N disks with 2 spare to replace any 2 that die in one go.
RAID 10 is a combination of RAID 0 and RAID 1. For specialists only who require the ultimate in uptime and speed (think datacentres) as it is expensive.

Think of a RAID "array" as a single disk as far as you are concerned (even if it has multiple disks in it) and you will be fine.

I remember a few years ago now experimenting with RAID, but you seem to have missed out on one which increased the size of the drive by combining two drives to make a larger one which was the sum of the two drives.

So you could RAID say a 250MB and a 500MB drive and get a 750MB drive.

Which RAID is that?
.
 
I remember a few years ago now experimenting with RAID, but you seem to have missed out on one which increased the size of the drive by combining two drives to make a larger one which was the sum of the two drives.

So you could RAID say a 250MB and a 500MB drive and get a 750MB drive.

Which RAID is that?
.

Sounds like just a bunch of disks to me ;)
 
Interesting/useful debate.
If you have two disks of the same size, you can use RAID0 to stripe data across them. So for example, 2 1TB disks would be paired in a RAID0 configuration to give 2TB of usable space. When a piece of data is written to this raid array, a chunk of it will be written to the first disk of the array and another chunk will be written to the second disk of the array.

This is different to JBOD where disks of different sizes can be used and their total storage capacity is amalgamated.

Currently I have:
> Random crap saved on PC, LRCatalogs backed up to...
> > NAS box (Raid 0), also home to actual images and business files
> > Static copy of NAS data on Fileserver, periodically/manually updated with RSync (RAID 0)
> > Static copy of NAS on external HDD

Right now I'm sorting out some issues with the Fileserver, it's nearly there. (Too many projects on the go!)
As soon as the Fileserver is working/tested I'll be using a set-up something along the following lines:
> Desktop PC with RAID1 disks - backed up frequently* to... (Desktop OS/Configuration periodically imaged to fileserver)
> > File server with RAID1 disks - incrementally** backed up to
> > > NAS with RAID0 disks (If I was still running a business, I'd invest in a 4 bay NAS with bigger disks and use them in RAID1 or RAID5 configuration)
> > > Something cloudy (Probably Amazon Glacier - a restore from here means poop has really hit the fan)

* I've yet to decide on an actual tool to implement this. I'm currently thinking about Unison
**Incremental back-ups provide a backup of files that have changed or are new since the last incremental backup. In the past I've used Acronis to do incremental backups from my Desktop PC - when things did go horribly wrong (thanks to a scuppered Windows Update) I was able to restore from the initial backup but the incremental backups failed.

Personally I tend to treat cloudy backups as tertiary. And quoting myself, external hard disks tend to be fragile. The disks themselves are easily damaged and the USB controllers seem to be generally quite naff.
I learned the lesson about using a second NAS for backup long ago after the power supply on my first NAS failed within days of installing it. I was unable to continue working until a replacement arrived. As photography was my business at the time, this had a severe impact.
 
Last edited:
what they said. if you raid disks of varying sizes, the larger ones would only have the same space as the smallest one available.

works the same with disk speed. if you raid multiple speed drives, they'll all throttle to the slowest drives speed.

But RAID0 is used to increase the apparent speed, is it not, of the array - 2 disks will double the speed 3 disks will increase the speed by 3x etc.

It doesn't increase the size of the array which defaults to the smallest disk size.
 
Last edited:
But RAID0 is used to increase the apparent speed, is it not, of the array - 2 disks will double the speed 3 disks will increase the speed by 3x etc.

it is used to increase speed. its probably the only exception to the rule. however i believe im right in saying it would only increase the speed based on the slowest disk multiplied by the amount of disks (and even then it would be a theoretical maximum) and not strictly a 2x, 3x etc speed increase.

It doesn't increase the size of the array which defaults to the smallest disk size.

dont understand what youre saying here.

e: i think i get what youre saying, raid0 is an exception to the smallest disk rule. i should have said that earlier.
 
Last edited:
Interesting/useful debate.
If you have two disks of the same size, you can use RAID0 to stripe data across them. So for example, 2 1TB disks would be paired in a RAID0 configuration to give 2TB of usable space. When a piece of data is written to this raid array, a chunk of it will be written to the first disk of the array and another chunk will be written to the second disk of the array.


And as I mentioned earlier, RAID 0 actually doubles your risk of hardware failure. If one of the drives fails you are guaranteed to lose the contents of both (unless some very exotic recovery saves you). From a data integrity POV you are better writing at random to one or the other.
 
RAID 5 is the thinking mans RAID - N disks, with 1 spare to replace any that fails. This is what most sensible people use at home.
RAID 6 is the rich thinking mans RAID - N disks with 2 spare to replace any 2 that die in one go.
Maybe a simplified explanation or choice of wording but for clarity there is no spare in RAID 5 or 6 by default, that requires an additional disk which is defined as a spare (or often called hot spare). In both R5 / R6 a parity block is written to each disk, one in the case of R5 and two for R6. This means a R5 array can sustain the loss of one drive across the array and remain functional. That failed disk needs to be replaced asap because if another disk fails the whole array will fail and you would lose all data. R5 requires a minimum of 3 disks to operate. In R6 two parity blocks are written to each disk and across the array it can sustain the loss of two disks but at the cost of requiring more disks to start with (4).

Useful tool to calculate redundancy, final space and disks required : http://www.raid-calculator.com/default.aspx

A RAID solution regardless of type is a backup if it does not contain the original, obvious really. Some forms of RAID improve speed while some mitigate disk failure but not deletion, virus infections etc.

Personally i do the following.

Originals are on a Server 2008R2 + RAID 6 + USB3 external drive (1 of 3 drives). My workflow is to copy files from cards to the server, then sync the copied files to the external storage. A second R2 box + RAID 5 then syncs the data across the lan (different location on the same site, different UPS / ring main covering it). The USB3 drive is periodically replaced with one that lives in a fireproof box and one that is taken off site on a rotation basis so in the event of catastrophic failure of the original site (e.g. fire, theft of 2 x servers + 2 usb) i still have accessible data off site.

I don't like the idea of cloud storage due to on-going costs, data security, upload speeds and download recovery time. In the event of a total loss of the original site i can plug in the offsite USB drive and decrypt it on any computer very quickly and efficiently.

/my 2p's worth
 
Last edited:
Maybe a simplified explanation or choice of wording
Simplified wording.

My media is stored on RAIDZ (RAID5 like array) with a hot swappable spare attached to the array ready.

I have 3 copies of data locally and cloud storage. I'm not too worried about download speeds from the cloud - if I'm having to recover from there a whole lot has gone wrong locally I'll be too busy worrying about that first....
 
2) if it can be a backup then how should it be used to do so properly?

A decent backup strategy will have your data safely stored in at least two places. A raid drive had two disks in it but it counts as a single place because the two disks always have the same info. A RAID drive can be used as part of a backup strategy - but it is treated as a single place. So you still need another place (an external hard drive, a NAS, the cloud, DVDs, etching onto granite blocks).

My backup is -

A NAS - two drives in a RAID setup, one place.
External hard drives - three of them, one attached to the computer, one stored next door, one stored off site.

So I have everything backed up to four places.
 
can you elaborate neil - what is raid and why is it not a good backup idea?

RAID merely gives redundancy: A drive fails, and you can carry one, fit a new drive and rebuild the RAID. It does not protect you from accidental deletion. It will also not protect you from file corruption should anything crash while writing to the RAID. All RAID1 will do is faithfully mirror that corrupted file to the other disk. RAID will also not protect you against theft, fire, flood etc. With RAID you still only have ONE copy of your files. Back up means you have the files on another system, preferably not even a second disk in the same system.

If you think you have back up because you're using RAID.... you'll regret it if someone breaks in and steals your computer one day, as you'll realise then that you didn't have back up at all.
 
RAID merely gives redundancy: A drive fails, and you can carry one, fit a new drive and rebuild the RAID. It does not protect you from accidental deletion. It will also not protect you from file corruption should anything crash while writing to the RAID. All RAID1 will do is faithfully mirror that corrupted file to the other disk. RAID will also not protect you against theft, fire, flood etc. With RAID you still only have ONE copy of your files. Back up means you have the files on another system, preferably not even a second disk in the same system.

If you think you have back up because you're using RAID.... you'll regret it if someone breaks in and steals your computer one day, as you'll realise then that you didn't have back up at all.

Yeah I think all of this was already covered above.

And also don't forget that raid CAN be a good backup as long as it isn't your primary source and it located at a separate location

Thanks
 
Last edited:
Yeah I think all of this was already covered above.

And also don't forget that raid CAN be a good backup as long as it isn't your primary source and it located at a separate location

Thanks


Yup. My back up servers are RAID.
 
Back
Top