Kevin Connery wrote:
My local data is in a raid-5. The local backup is also raid-5. Offsite is also raid-5.

6 drives would have to fail, be stolen, broken, etc before the data's lost.

(Yes, I'm paranoid.)

actually, one controller, a few drives, corrupt data and you are toast. most people forget about the controller board and/or data corruption. you can suddenly have a cluster of fully working drives and no way to get at the data. it's why enterprise servers have duplicate controllers and enterprise NAS have dual ethernet ports. AND they still have backups done.

http://www.adrc.com/raid_failure_types.html

one guess, what happens when you can't find a matching, compatible controller? a power surge or hiccup may corrupt the RAID configuration setting of NVRAM in the controller card. of course, if you are using a system board controller, it could be worse.

for fire and floods there is iosafe, expensive, heavy:
https://iosafe.com/products-5baynas-overview

Kevin Connery wrote:
My local data is in a raid-5. The local backup is also raid-5. Offsite is also raid-5.

6 drives would have to fail, be stolen, broken, etc before the data's lost.

(Yes, I'm paranoid.)

Not paranoid, extra cautious.

Kevin Connery wrote:
My local data is in a raid-5. The local backup is also raid-5. Offsite is also raid-5.

6 drives would have to fail, be stolen, broken, etc before the data's lost.

(Yes, I'm paranoid.)

Leonard Gee Photography wrote:
actually, one controller, a few drives, corrupt data and you are toast. most people forget about the controller board and/or data corruption. you can suddenly have a cluster of fully working drives and no way to get at the data. it's why enterprise servers have duplicate controllers and enterprise NAS have dual ethernet ports. AND they still have backups done.

http://www.adrc.com/raid_failure_types.html

one guess, what happens when you can't find a matching, compatible controller? a power surge or hiccup may corrupt the RAID configuration setting of NVRAM in the controller card. of course, if you are using a system board controller, it could be worse.

for fire and floods there is iosafe, expensive, heavy:
https://iosafe.com/products-5baynas-overview

The main data is stored locally on a raid from OWC. I've had one drive die and replaced it.

The local and offsite backups are on Synology DS512+ NAS systems--the previous year's model to the iosafe service uses. Deliberately not the same system as the main one. I'd considered using 2 different backup options to cycle between, but decided the complexity wasn't worth the potential benefit. The two backups, though, have different models of drives.

However you look at it, all 3 systems have to fail in a very short period of time before the frantic phone call to a recovery service.

A couple of EMP pulses and they could all be destroyed in seconds; my offsite storage is less than 10 miles from home. There's a limit to how far out on the bell curve I'm willing to protect against, though.

Armando D Photography wrote:
That was some great timing. I just bought an single external drive to backup my raid. Then just yesterday it failed (2 drives). Those times I've read on the forum about people having even 3 backups. It's true, and it can happen. You can't just have 1 backup, it really does have to be 2 backups minimum. Using recuva to see what I can bring back, but happy it's on another drive.

Why post this? It happens.

also, you should really get OFFSITE backup. preferably in another part of the country.

PhotoByWayne wrote:
How many hours had those 2 failed drives ran for?

for 3 years, hgst

wolfdenlab wrote:
Have to agree with a few of these tips. I'm doing the same. I've lost crucial files too many times to not protect myself and data. (I'm also an IT guy by trade)

I have a RAID 10 array with all of my shoots on it. (Good idea to mix up the makes and models of drives.) I then auto-copy (Robocopy) that onto a secondary internal drive, and then onto my NAS.  Then I have it sent to my cloud drive (Carbonite). 

Kudos for good, helpful advice.

I'm curious why you think it's a good idea to mix up the drives?

+1 on cloud storage for offsite backup.

Cheers!
Noah

Noah Russell wrote:

I'm curious why you think it's a good idea to mix up the drives?

+1 on cloud storage for offsite backup.

Cheers!
Noah

A good backup strategy minimizes single points of failure that can destroy all copies of your data.     For instance, if all copies are in the same room, a fire can easily destroy all copies.  Similarly, if your data only exists on a single RAID device, then a failure of the RAID controller can destroy all copies.

The theory behind using different brands of hard drive is to minimize the possibility that a manufacturing problem can destroy your data.

In the past, there have been hard drive issues with a bad batch of lubricant.  After a few years, these drives would fail to start up.   If all your drives are from the same manufacturing batch, problems with lubricants or other manufacturing issues can cause all your drives to fail at about the same time.  Similarly, a firmware problem could also cause drives to fail at a certain age.

By using drives from multiple manufacturers, you decrease the chances that a single manufacturing defect can affect all your drives.

Thise of us who are really paranoid also keep multiple copies on different types media.  Perhaps a copy on traditional HD, and a copy of burnable Blu_Ray.  Copies on optical discs are not easily susceptible to damage from water or magnetism.   

Anything you can do to eliminate common failures, will increase the reliability of your backups.

Michael Fryd wrote:

A good backup strategy minimizes single points of failure that can destroy all copies of your data.     For instance, if all copies are in the same room, a fire can easily destroy all copies.  Similarly, if your data only exists on a single RAID device, then a failure of the RAID controller can destroy all copies.

The theory behind using different brands of hard drive is to minimize the possibility that a manufacturing problem can destroy your data.

In the past, there have been hard drive issues with a bad batch of lubricant.  After a few years, these drives would fail to start up.   If all your drives are from the same manufacturing batch, problems with lubricants or other manufacturing issues can cause all your drives to fail at about the same time.  Similarly, a firmware problem could also cause drives to fail at a certain age.

By using drives from multiple manufacturers, you decrease the chances that a single manufacturing defect can affect all your drives.

Thise of us who are really paranoid also keep multiple copies on different types media.  Perhaps a copy on traditional HD, and a copy of burnable Blu_Ray.  Copies on optical discs are not easily susceptible to damage from water or magnetism.   

Anything you can do to eliminate common failures, will increase the reliability of your backups.

I get it, but a fault tolerant raid array is not a backup. You still have only one copy of your data and it's all in one place. The file system/files could still be corrupted or inadvertently or maliciously erased. The raid controller itself can also conspire against you.

Cheers!
Noah

Noah Russell wrote:

I get it, but a fault tolerant raid array is not a backup. You still have only one copy of your data and it's all in one place. The file system/files could still be corrupted or inadvertently or maliciously erased. The raid controller itself can also conspire against you.

Cheers!
Noah

Absolutely.  Store your data in a RAID box, and you have not eliminated the need for a backup outside the RAID.

My point was that if you have a 2 drive mirrored RAID, there is less chance for both drives to fail at the same time if the two drives are from different manufacturers.

Furthermore, you get a higher degree of independence if your master copy is on one brand of drive, and your backup copy is on a different brand of drive.

The difference may be small, but this is the sort of thing that should be considered when bacing up important data.

Noah Russell wrote:
It may seem like I'm arguing semantics, but raid and backups are two completely different and unrelated things.

Using different types of media and storing them in different locations are relevant to a backup strategy.  Using drives from multiple manufacturers in your local raid set is not.

...

I don't think it's semantics.  I think we disagree on some aspects of the issue.

We both agree that "RAID" and "backups" are different. 

One of the advantages of a RAID is that it can be used to increase reliability.   Many RAID configurations can tolerate the failure of a single drive mechanism, without a disruption in service.  However, if a second drive fails before you replace the first, the data on the RAID is lost, and you need to restore from a backup.   

In many situations, there are significant advantages to minimizing downtime. 


In a two drive RAID (where each drive is a mirror of the other), you want to minimize the chance that both drives will fail at the same time.  Using drives from different manufacturers reduces the chance that a second drive will fail before the first is replaced.

I agree that this does not eliminate the need for backups, but it reduces the likelihood that you will need to restore from the backup.


The above is a broad generalization.  There are RAID-like devices (like the DROBO) that will automatically redistribute data when one drive fails, to prepare itself in case a second drive fails.   There are also multi-drive configurations which can continue operation with two failed drives.

Michael Fryd wrote:

I don't think it's semantics.  I think we disagree on some aspects of the issue.

We both agree that "RAID" and "backups" are different. 

One of the advantages of a RAID is that it can be used to increase reliability.   Many RAID configurations can tolerate the failure of a single drive mechanism, without a disruption in service.  However, if a second drive fails before you replace the first, the data on the RAID is lost, and you need to restore from a backup.   

In many situations, there are significant advantages to minimizing downtime. 


In a two drive RAID (where each drive is a mirror of the other), you want to minimize the chance that both drives will fail at the same time.  Using drives from different manufacturers reduces the chance that a second drive will fail before the first is replaced.

I agree that this does not eliminate the need for backups, but it reduces the likelihood that you will need to restore from the backup.


The above is a broad generalization.  There are RAID-like devices (like the DROBO) that will automatically redistribute data when one drive fails, to prepare itself in case a second drive fails.   There are also multi-drive configurations which can continue operation with two failed drives.

I don't think we disagree then. I should have just read the whole thread because the point I was making had been made before by several other folks.

raid helps you stay online by providing 'fault recovery.' I say fault recovery because a single drive failure is a pretty minor event for an 'enterprise' class server. Just pull out the failed drive and put in a good one and in a couple hours everything is back to normal. No tools or reboot/downtime required and it probably sent someone telling them to replace the drive.

Backups provide disaster recovery allowing you to recover from catastrophic events. Godzilla is coming, we must flee the city! You sank my battleship type stuff. 

For example:

http://royal.pingdom.com/2008/07/18/for … et-stolen/

http://www.engadget.com/2014/04/20/sams … -sds-fire/

Alexxiss wrote:
Having been running the same 6 drive RAID-5 array for years with identical drives, I have had two drives fail nowhere near at the same time.

You probably have not had the RAID stolen, nor a roof leak.

Reliability and backups are about anticipating problems and preparing for them.   Some problems are likely to happen (drive mechanism failure), and some are rare (building burning down).   

There are lots of rare failure modes.  Enough of them that you shouldn't be surprised if you run into one of them.

In the past there have been manufacturing issues that caused drives from the same batch to fail at about the same time.  Buying drives from different manufacturers is one way to reduce the chance you will see multiple failures at about the same time.



I am not suggesting that using oe brand of drive is a recipe for disaster, nor am I suggesting that different brands of drive make a RAID 100% reliable.   I am merely providing an example of the types of issues that should be considered when devising a strategy for preserving very important data.

Michael Fryd wrote:
You probably have not had the RAID stolen, nor a roof leak.

Actually I have had the roof leak, though not in the same room.  But even if it did leak in this room it is unlikely to get to the RAID array being at the bottom of the 44u rackmount cabinet.

Michael Fryd wrote:
In the past there have been manufacturing issues that caused drives from the same batch to fail at about the same time.  Buying drives from different manufacturers is one way to reduce the chance you will see multiple failures at about the same time.

I am not suggesting that using oe brand of drive is a recipe for disaster, nor am I suggesting that different brands of drive make a RAID 100% reliable.   I am merely providing an example of the types of issues that should be considered when devising a strategy for preserving very important data.

Yes, it is a possibility but not an automatic disaster or anything.

Was just showing it doesn't have to be a problem.

Note that RAID array is the backup that is not the only place stuff is stored.