Why Implement RAID for Backup?

I have seen MANY posts here promoting RAID as a backup solution for media libraries.

Maybe you guys could explain why.

I could be wrong but to me RAID is a redundancy solution not a backup system.

If my library was to crash. RAID would certainly help me get back up pretty fast.

However, aren’t all the drives in a RAID array all pretty much in the same physical location?? Sooo… if there was a disaster that destroyed the drives in a single location, would not that make RAID useless.

I maintain 3 physical copies of my library in different locations, one being portable drives for offsite backup… (The portable copy can travel with me and I have access to my entire library on the road without using streaming data over my cell network.)

I am in Southern California so fires and earthquakes happen, so if my main library was destroyed I have a complete backup. (Although it would take more time to recover, at least I stand a better chance of my library surviving.)

What are your thoughts??

Thanks

You are exactly right. RAID is not a backup and should never be considered as such. Anyone that promotes otherwise will sorely be mistaken if something happens to the main system as you pointed out.

Allot of people don’t have the foresight or concern you are discussing to them. They simply think in the terms of a drive failure which raid or any of the drive duplication technologies can protect against.

I agree that Raid is not a backup solution. The reason I cannot stretch my belief to think of Raid as a backup has already been mentioned, it does not protect against some single point failures. That is should the Raid board fail you loose access to all file that are controlled by that Raid system.

There is something like a Raid system that I do think of as a basic backup and that is StableBit’s DrivePool. DrivePool is different in that it keeps all files in a pool in a standard Windows format and they can be read completely outside the DrivePool system you just need to hook the drive to any windows computer and navigate to the directory DrivePool created. Along with that DirvePool assures that, with duplication turned on, files are always kept on different drives in the pool.

DrivePool is not a fully functional backup because there is no provision for detaching a drive and storing it in a different location and, unless you turn on 3X or more duplication, there is no way to be sure of recovering from two drive failures at the same time.

Ideally any pooling solution should include an external backup solution but I have been running DrivePool for some time and have had a couple of drive failures and one full computer failure and I have lost no content. So I do consider DrivePool, properly configured, to be at least a reasonable local backup solution.

@Elijah_Baley said:

So I do consider DrivePool, properly configured, to be at least a reasonable local backup solution.

Until your house burns down, of your server gets stolen, or you spill a jug of water on it, or your filesystem corrupts, or malware encrypts it, or your kids delete it.

Redundancy is not backup.

@trudge said:

@Elijah_Baley said:

So I do consider DrivePool, properly configured, to be at least a reasonable local backup solution.

Until your house burns down, of your server gets stolen, or you spill a jug of water on it, or your filesystem corrupts, or malware encrypts it

Redundancy is not backup.

Which is exactly why I said “local backup” and why I also said “DrivePool is not a fully functional backup because there is no provision for detaching a drive and storing it in a different location and, unless you turn on 3X or more duplication, there is no way to be sure of recovering from two drive failures at the same time.”

It is amazing what you can “prove” by picking a small part of a longer statement and pulling it out of context.

If any of your stated “disasters” should happen then my problems are much greater than just recreating my local media and I do not plan for extreme cases. I believe in the formula of “Plan for the average and then be ready to cope with the abnormal.” I can recreate most of my entire library from scratch from DVDs or VHS tapes that are stored elsewhere. I do not really consider that a “backup” either because of the trouble it would entail to do the recreation.

DrivePool is, for me, a sufficient local backup solution and that is all I need. After all it is just entertainment and there is a limit to what I am willing to pay to protect it.

A sort of an expensive false security I believe…

@Elijah_Baley said:

@trudge said:

@Elijah_Baley said:

So I do consider DrivePool, properly configured, to be at least a reasonable local backup solution.

Until your house burns down, of your server gets stolen, or you spill a jug of water on it, or your filesystem corrupts, or malware encrypts it

Redundancy is not backup.

Which is exactly why I said “local backup” and why I also said “DrivePool is not a fully functional backup because there is no provision for detaching a drive and storing it in a different location and, unless you turn on 3X or more duplication, there is no way to be sure of recovering from two drive failures at the same time.”

It is not a local backup, it is disk redundancy. A local backup implies you can survive a server localised disaster. I’ve just given you multiple examples (and far from a complete list) of server localised disasters your data cannot survive.

If you setup is sufficient for you then that is fine, but don’t redefine existing terms by calling it something it isn’t - it just confuses people who don’t know any better.

@trudge said:

@Elijah_Baley said:

@trudge said:

@Elijah_Baley said:

So I do consider DrivePool, properly configured, to be at least a reasonable local backup solution.

Until your house burns down, of your server gets stolen, or you spill a jug of water on it, or your filesystem corrupts, or malware encrypts it

Redundancy is not backup.

Which is exactly why I said “local backup” and why I also said “DrivePool is not a fully functional backup because there is no provision for detaching a drive and storing it in a different location and, unless you turn on 3X or more duplication, there is no way to be sure of recovering from two drive failures at the same time.”

It is not a local backup, it is disk redundancy. A local backup implies you can survive a server localised disaster. I’ve just given you multiple examples (and far from a complete list) of server localised disasters your data cannot survive.

If you setup is sufficient for you then that is fine, but don’t redefine existing terms by calling it something it isn’t - it just confused people who don’t know any better.

Backup as used here from the Merriam-Webster dictionary

3: a copy of computer data (as a file or the contents of a hard drive); also : the act or an instance of making a backup

Redundancy: the closest definition I can find to what is meant here from the same source:

3: serving as a duplicate for preventing failure of an entire system (as a spacecraft) upon failure of a single component

So it seems that both words are similar in meaning and use but only “backup” refers to computer data directly. You may define them differently and even use them differently BUT under the definition of “backup” a local copy IS a backup. It is not a 100% safe backup but it is a backup. Planing for the disasters you mentioned is, in many cases, just not needed as what you mention are so farfetched or so serious that planing for them and spending money on them is absurd for many people.

In the first case (like theft or spilling water on my server) those are so unlikely as to be absurd to consider. My drives are external to my computer so the likelihood of two drives corrupting or failing at the same time for any reason is small and my server is not going to get malware except by an extreme fluke.

Backup is NOT being perfectly safe it is protecting your stuff to the degree you are willing to live with. DrivePool, for me, provides that protection so I consider it a local backup. It could even be safer as it could use the cloud for the duplicated but my upload is so slow that it would take a year or better to get everything uploaded.

You can define “backup” however you want but I choose to use the term as it is defined in common English so DrivePool is a local backup solution for me.

You are also in error as to what “local backup” means as “local backup” just means the ability to recover from failures using local resources not the ability to recover from things that destroy what is local.

I have the ability to recover locally from any reasonable failure that can happen so I have a reasonably good local backup. If anything worse happens, and it is possible just unlikely, I will simply cope with it. As I said before it is only entertainment and not worthy of over thinking it or over spending on it.

Aw I always come to the fun conversations late and my like-minded buddies beat me to saying all the important stuff.

Yes, RAID is not “backup”. It’s for uptime/availability. Hard drive failures are annoying inevitabilities and it’s a question of how much of a PITA you want to deal with when (not “if”) one happens. We do love our Plex server and it gets a lot of use, plus it’d kind of suck rebuilding everything. So RAID is some insurance so that when (not “if”) a drive fails it’s really not a big deal. Now, if the computer gets stolen or the house burns down, we have bigger issues to deal with. But there’s nothing on our NAS RAID array that can’t be replaced since its solely used for Plex so it’s all just ripped movies. If we were using the NAS for other stuff as well then an actual backup solution would be in order.

You do not think RAID is over spending??

With RAID you still have the same storage requirement for redundancy (same # of HDDs)
but not the security of a safe backup copy.
Discrete backup HDDs also offer a level of simplicity and reliability not found in RAID solutions.

Cost vs. Benefit does not seem equitable to me.

Just my opinion, as my background was spent a lot in disaster recovery planning for major banks.
Occupational hazard I guess … LOL

I do suppose you could make a discrete copy of your RAID array for separate storage. But how much extra will that cost??

And yes it is just entertainment (LOL we are not saving lives here…) but many of us, you included, have many MANY hours invested in our PMS and Library

I, for one, would not want to start from scratch again… (LOL Prolly would never be able to find some of these DVDs again! My movie library spans nearly 100 years of film making and my TV library spans decades of classic TV)

RAID is most certainly not a backup solution of any kind, it’s just redundancy for uptime. The word backup shouldn’t even be mentioned. If you delete a file off of your RAID array, it’s gone. Disk redundancy isn’t going to help you retrieve that file.

@Elijah_Baley said:

@trudge said:

@Elijah_Baley said:

@trudge said:

@Elijah_Baley said:

So I do consider DrivePool, properly configured, to be at least a reasonable local backup solution.

Until your house burns down, of your server gets stolen, or you spill a jug of water on it, or your filesystem corrupts, or malware encrypts it

Redundancy is not backup.

Which is exactly why I said “local backup” and why I also said “DrivePool is not a fully functional backup because there is no provision for detaching a drive and storing it in a different location and, unless you turn on 3X or more duplication, there is no way to be sure of recovering from two drive failures at the same time.”

It is not a local backup, it is disk redundancy. A local backup implies you can survive a server localised disaster. I’ve just given you multiple examples (and far from a complete list) of server localised disasters your data cannot survive.

If you setup is sufficient for you then that is fine, but don’t redefine existing terms by calling it something it isn’t - it just confused people who don’t know any better.

Backup as used here from the Merriam-Webster dictionary

3: a copy of computer data (as a file or the contents of a hard drive); also : the act or an instance of making a backup

Redundancy: the closest definition I can find to what is meant here from the same source:

3: serving as a duplicate for preventing failure of an entire system (as a spacecraft) upon failure of a single component

If you can’t see the difference between both those terms and which matches drivepool functionality then there is zero point going further.

If I turned up at your house tomorrow and smashed 1 of your drivepool drives with a big hammer, which term fits your setup? Both right?

If I turned up at your house tomorrow and smashed 2 of your drivepool drives with a big hammer, which term fits your setup? Just redundancy right?

So what setup do you actually have?

There is nothing about the likelihood of a disaster in either term. It doesn’t matter if there is a 0.00000000001% chance of some nasty bit rot corrupting some of your files tomorrow, it doesn’t change the definitions you have posted.

To toss a little extra into the stew :wink:

Here is how I do NAS “backups”.

  1. 25, soon to be 36, TB using external HD in their own enclosure as shown below.
  2. NAS does a ‘rsync’ (recursive copy/sync), preserving structure, name, date/time stamp. It does not automatically span HDs (I do that)
  3. As you can see, 5 enclusures are active. “6” is sitting off to the right about to go into service.

These drives, when turned on, are accessible through the NAS to my desktop (Linux) at 100 MB/s. It writes directly to the backup drive and doesn’t clutter the NAS.

Apologies for the clutter. New 4K monitor (PB287Q) arrived 2 days ago. Still getting things put back together

@trudge said:

So what setup do you actually have?

My backup solution is DrivePool with duplication for local backup and offsite original copies for 90% of my current library. I also have some files, those that I consider important enough to allow my computer to spend the time to upload, stored in the cloud on Amazon Drive.

I feel safe enough and I am not going to protect against extreme long shots any more than I will play the lottery in the hopes of getting rich. I also do not carry specialty insurance. If disasters happen I will just deal with them. I do not throw money or time at low probability risks.

And yes I clearly know the difference between backup and redundancy. I just consider them mostly the same. Standard Raid is on the redundancy side of the line because a single point of failure (raid controller) can render everything lost while DrivePool is on the backup side because no reasonable single point failure will cause anything to be lost.

@Elijah_Baley said:

@trudge said:

So what setup do you actually have?

And yes I clearly know the difference between backup and redundancy. I just consider them mostly the same. Standard Raid is on the redundancy side of the line because a single point of failure (raid controller) can render everything lost while DrivePool is on the backup side because no reasonable single point failure will cause anything to be lost.

There you go modifying the definitions you posted just so they fit your point of view.

The world single applies to redundancy by the very definition you want to go by. And reasonable is your own interpretation of likelihood which apples to neither.

You can refer to them by any means as you want, and run your system however you want, all I am trying to say is please stop confusing people who know less by incorrectly using them in discussions such as this.

Here’s my example of RAID + Backups.

Server at the Bottom above my UPS (UnRAID01) is my main array (64TB) which is essentially a RAID6 setup (UnRAID dual parity). Server above that (UnRAID02) is essentially the same except for the fact it’s currently only using single parity (so essentially RAID5). The one major difference between these arrays being “like” RAID5/6 and actually RAID5/6 is that the data is not striped across the disks so losing an additional disk beyond parity does not render the rest of my data unrecoverable.

UnRAID01 does an rsync to UnRAID02 twice a day for redundancy. So there you have it. RAID + redundancy for both maximum uptime and local backups.

Note I also backup my important data to an offsite (parent’s house) RAID1 array (Synology) over site-to-site VPN.

@jjrjr1 said:
You do not think RAID is over spending??
Just my opinion, as my background was spent a lot in disaster recovery planning for major banks.
Occupational hazard I guess … LOL
And when doing that, when with any data deemed critical did you go “no, we cant protect that as it’ll be too costly?”

This isn’t a “having to keep up with the Jones’s” type issue, know your technology and options and apply what fits best for you. Just cause others do something doesn’t mean you have to.

I run Array 1 as a primary, with hourly Array snapshots.
Array 2, which every day takes and rsync copy of Array 1
Array 1 also runs a continuous backup to a cloud provider.

I choose to do this as I view my time and investment to create the data as more valuable to me than the cost of maintaining it.

@trudge said:

I choose to do this as I view my time and investment to create the data as more valuable to me than the cost of maintaining it.

Amen.

@trudge said:

@Elijah_Baley said:

@trudge said:

So what setup do you actually have?

And yes I clearly know the difference between backup and redundancy. I just consider them mostly the same. Standard Raid is on the redundancy side of the line because a single point of failure (raid controller) can render everything lost while DrivePool is on the backup side because no reasonable single point failure will cause anything to be lost.

There you go modifying the definitions you posted just so they fit your point of view.

The world single applies to redundancy by the very definition you want to go by. And reasonable is your own interpretation of likelihood which apples to neither.

You can refer to them by any means as you want, and run your system however you want, all I am trying to say is please stop confusing people who know less by incorrectly using them.

At NO point in the definition of “backup” does it say that the copy made must be off site or even on a different device. It simply says that a backup is a copy so it is you that is making it into what you want it to be.

My media files ARE backed up and I will continue to use the correct definition of “backup” in my posts.

If you want to make a distinction then rate the various forms of backup from unsafe/weak to safe/strong (from 1-10) but saying that a copy made by DrivePool and even by standard Raid is not a backup is just changing the language to fit what you want to believe.

I guess I will never convince anyone to change their views so I will use “backup” the way I believe is correct and others can use it as they believe is correct but furthering this debate is just a waste of time as it is clear that no one is going to change so I will move on and try, maybe, to do something that is actually productive and maybe even meaningful.

@Elijah_Baley said:

@trudge said:

@Elijah_Baley said:

@trudge said:

So what setup do you actually have?

And yes I clearly know the difference between backup and redundancy. I just consider them mostly the same. Standard Raid is on the redundancy side of the line because a single point of failure (raid controller) can render everything lost while DrivePool is on the backup side because no reasonable single point failure will cause anything to be lost.

There you go modifying the definitions you posted just so they fit your point of view.

The world single applies to redundancy by the very definition you want to go by. And reasonable is your own interpretation of likelihood which apples to neither.

You can refer to them by any means as you want, and run your system however you want, all I am trying to say is please stop confusing people who know less by incorrectly using them.

At NO point in the definition of “backup” does it say that the copy made must be off site or even on a different device. It simply says that a backup is a copy so it is you that is making it into what you want it to be.

My media files ARE backed up and I will continue to use the correct definition of “backup” in my posts.

If you want to make a distinction then rate the various forms of backup from unsafe/weak to safe/strong (from 1-10) but saying that a copy made by DrivePool and even by standard Raid is not a backup is just changing the language to fit what you want to believe.

I guess I will never convince anyone to change their views so I will use “backup” the way I believe is correct and others can use it as they believe is correct but furthering this debate is just a waste of time as it is clear that no one is going to change so I will move on and try, maybe, to do something that is actually productive and maybe even meaningful.

So if you mis-click and delete a file, it’s backed up? I think not.