RAID is not a backup.

[WoodsDweller]

While reading reviews today I came upon this comment by some user, dated April 2008. I'm not going to identify the manufacturer of the NAS unit, because it doesn't matter. I'm not going to link to the original post, because it identifies the manufacturer, and it doesn't matter. It has nothing to do with any shortcoming on the part of the manufacturer, it is not a problem with their tech support (contrary to what the original poster seems to think), and only somewhat on the part of the user (he should have backed up his RAID array before applying a firmware upgrade).

 

When I mention (repeatedly, and people may be getting tired of it by now) that RAID isn't a backup, this is one example of the sort of thing I'm talking about.

 

Stuff happens. Your data, and the time you spend producing your data, are important, too important to be working without a net. Back it up!

 

 

After applying the first firmware update, the *** rebooted and my Mac informed me that the connected disk was unformatted and asked what I wanted to do (Initialize, Ignore, Eject). As you can imagine, I was immediately filled with panic and dread.

 

Over the past five years or more, I've built a huge (legal) music collection. I'm a music freak and buy between 5 and 10 CDs each month. I had painstakingly ripped each and every one of my 2000+ CDs to the ***, both in Apple Lossless format for my home stereo and AAC format for my iPod. I can't even imagine how many hours (hundreds?) I spent doing this. I also took the time to verify all of the metadata (artist, album, etc) and correct any tag errors. As you can imagine, this was VERY TEDIOUS!!!

 

Recently I've been ripping my DVD collection and converting it to MP4 for my Apple TV. Each rip/conversion takes between 3-4 hours. A hundred DVDs or so and you're looking at another 300 hours of time spent.

 

So, I called *** tech support. They walked me through a procedure that involved removing and re-inserting the drives. At first this didn't work. Then the tech told me to change the position of the drives in the ***. Voila! It took about three days to "protect" the data. After this process had finished, my data was back and I was much relieved.

 

Until the next firmware update.

 

The most recent update completely hosed my data. This time, tech support's procedure didn't work. It took 12 days to "protect" my data and when all was finished, my data was lost. Throughout this experience, I made efforts to contact tech support. I sent them my ***'s diagnostic files, as requested. Someone from *** sent me the occasional terse email saying "This is a high priority case!" or "Please be patient!" A month later, how patient am I supposed to be???

 

At this point, I've accepted that my [2 TB of] data is gone.

 

 

[The Computer Audiophile]

Great, but unfortunate, example WoodsDweller.

 

Thanks for all the detail.

 

[tmesselt]

Sorry to hear that, I would be very frustrated.

 

[Steve_S]

Yes and no. There are various forms of Raid, all for different purposes. Raid-1 is the most common for data security, and involves two hard drives, each reading and writing identical data. If one hard drive fails and the data is lost, the system will switch over to the other hard drive and continue uninterrupted, except for a warning about the hardware failure. So Raid-1 is indeed a backup against data loss.

 

Where Raid-1 will not protect you is against viruses, malware, accidental deletions, OS malfunctions, hackers or any other software-related data loss. Because each bit written to one drive is also written to the other, they act as a single drive and any changes or software-related damage to the data integrity will be shared by both drives. It may also fail to protect against power spikes or physical damage. The only way to protect against this type of loss is to keep a physically separate backup, such as external hard drive, CD, DVD, etc.

 

One could argue that these aren't really a secure backup either because there is the possibility of theft or fire, which would wipe out the original drive as well as the backup. So in this view we could determine that the only true backup is one kept in a different location from the original, and in fact this is the best way although not very convenient since you have to visit this location whenever you want to make a backup. This means backups will not be performed as often as they should which defeats the point of backing up at all! Then there are web-related backups but these aren't really secure either, so you might not want to store sensitive data in such a manner.

 

In the end, we decide what works best for our use, from type of data to the required level of security to convenience of backup.

 

[PeterSt]

Raid-1 is the most common for data security, and involves two hard drives, each reading and writing identical data.

 

Nitpicking would be : no, reading is not the same from each drive. In a good implementation elevator seeking is in order (as "invented" by Novell), which means the drive with the head closest to the data, provides it.

 

A little less nitpicking would be : no, RAID1 it is not the most common. RAID5 is.

 

The saddest of all is that RAID1 in hardware implementation (meaning separate controllers as well) has become quite undoable, because the backplanes needed for it can't be bought anymore. For some years they were still produced no reseller knowing about it, but now they have disappeared really.

 

So in this view we could determine that the only true backup is one kept in a different location from the original, and in fact this is the best way although not very convenient since you have to visit this location whenever you want to make a backup.

 

Assuming most people have jobs and those jobs are in an office outside of the house, it is easy to bring the backups to the office. Just like "normal life" goes the other way around : someone from the office takes the daily backup to his house.

 

The problem with backups of music (or movie) data is different though;

 

There is no way the data even can be backuped, just because the amounts are too large and even the fastest means (which would be a SATAII disk-disk copy these days) are way too slow. What I do is this (and I know, it is only the best of worse) :

 

I have 9 online 1TB disks and one of them is the current backup drive. When the backup disk is full, it is switched with another. The 8 live disks are not all completely full, but contain groupings of music (like one for classical, one for rock etc.).

Once per x days, groups of newly obtained albums go to the active backup disk. I administer this at the group level.

I never throw away anything, because it would imply to look up and mount the backup disk concerned, which is too tedious and besides implies mangling the backup disks.

When a backup disk is full, it is stored away and never touched again (awaitening restore activities which would be quite tedious in my case). Note that the backup disks are smaller on purpose (besides that it's cheaper), and during the process of filling one up, I'd loose both live and backup data from that one backup disk when the house burns down. Of course this can be arranged for by making a regular bakup of the active backup disk.

 

I think this is the only means which makes life liveable, and the only real downside is that you officially can't delete. Oh, of course you can if it is administered as well, but by that time I wouldn't begin a restore when it's needed.

 

I would never in my life use any form of RAID, because I'd just challenge for problems when whatever part runs out of order. The only reason would be a kind of "keep on running" at disk faliure, which is no reason at all for this application, when compared with the downsides.

This way I can switch disks how I want and to where I want (other PCs) and by which means I want (directly SATAII, eSATA, USB, Firewire) and I'm super flexible. I'm doing this for two years now, and I'm very satisfied with it. Even used the restore procedure for one of the data disks once ...

 

Peter

 

PS: Of course I don't treat the OS disk like this.

PPS: Don't underestimate the importance of a speedy connection with these amounts of data; where SATAII allows a 1TB copy in 90 minutes, with USB 2.0 this already comes down to a small day.

 

 

[The Computer Audiophile]

"Nitpicking"

 

Ah yes, it's nice to see our favorite nitpicker back in action here at CA :~)

 

I'd rather spend more time listening to music than messing with computer and disk configurations. Configs are very important and must be done correctly but "correctly" is different for everyone obviously. Since time is a finite resource it is definitely an either or situation. Either spend more time listening or spend more time as an IT admin. I use RAID 5 and think this is an acceptable level of risk for myself.

 

[Steve_S]

Nitpicky indeed! I will admit I misspoke when I said read and write, what I was trying to say is that the data on both drives is identical.

 

Since we're talking about home computers here, I will argue that RAID-1 is still the most common, although I have not seen any data (nor would it be accurate) to show what people are most commonly using today. It is not only very common but also fully supported in home computing, and will be for some time.

 

"Assuming most people have jobs and those jobs are in an office outside of the house, it is easy to bring the backups to the office."

 

Do you honestly feel this is not less convenient than doing a backup to an external drive hardwired to the machine being backed up? I disagree completely! As I stated, it is far more secure against theft or fire, but it is not convenient compared to the aforementioned.

 

"I would never in my life use any form of RAID, because I'd just challenge for problems when whatever part runs out of order. The only reason would be a kind of "keep on running" at disk faliure, which is no reason at all for this application, when compared with the downsides."

 

I'm curious what your "downsides" are. Having a mirrored RAID array means you will not lose any data in the event of drive failure. Scheduled or manual backups will not protect new data from being lost! It's very easy to set up, even for a novice user and costs only one hard drive.

 

"Either spend more time listening or spend more time as an IT admin."

 

Just for interest, I had a hard drive suddenly fail six weeks ago. RAID-1 was set up on that machine which means I continued working uninterrupted and replaced the disc when I had the chance. It was nice to spend less time dealing with hardware failure and more time listening to music.

 

It takes an extra ten minutes to set up a RAID configuration when setting up a new computer. I hope people can tear themselves away from music for that long!

 

[The Computer Audiophile]

I agree Steve. Setup RAID1 or RAID5 and get on to listening.

 

[PeterSt]

Haha, I was wrong at saying that I wouldn't use "any form of RAID". RAID1 is the best IMHO and nothing can go wrong with it. No hussle with exchanging disks either. Professionally we have been provocing this for 22 years now for all our ERP users (Novell's Duplexing at the time -> not Mirrorring) but today it has died out because of the lacking backplanes. That is, if you want to do it right with double controllers.

 

It also may be true that at homes RAID1 is more common than RAID5, but you must wonder whether this is usd "at all". I mean, relatively of course. People will be having enough problems mounting their "native" disks, let alone double them for this reason which ... is no reason at all.

 

Chris, this is not much subjective I think. If you say you take your chances on loosing your data, you may have a 100 CDs only in there. Otherwise it is, well ... nah, I say nothing.

1000s of hours go into ripping and all, and if I'd take my chances on that ... I'd rather jump of a bridge when it happens.

 

Let me try to put it differently :

All the companies I know use a form of RAID. All those same comanies make a daily backup as well.

Don't try to talk them out of either halve. This is not about taking chances.

 

RAID is not a backup.

Disks break.

 

[WoodsDweller]

A business has a constant stream of transaction data which needs to be protected, and they can't afford to lose the last 30 minutes of transactions. They also have reliability requirements, since a down system means they aren't making money.

 

But we're just talking about your music collection and the effort you put forth to rip it, and the backup requirements are a lot less rigorous. If you have 2000 discs ripped and you back them up, and you buy and rip 25 new discs over three months, and you lose your data and have to restore from a three-month-old backup, oh noes! you have to rip 25 discs over again. That's an inconvenience that you have to weigh against the time demands of more rigorous backups.

 

This shouldn't be presented as a false choice between not doing backups, vs. painstakingly doing daily off-site backups.

 

Your data is far safer if stored on one single drive with a periodic copy onto a second, external drive, than it is with a RAID 1 using two discs and no backups. Of course, that's another false choice, since you could, for example, store on a RAID array, and have a backup on a single drive (if it's big enough), or a second RAID array.

 

RAID isn't a problem, feeling safe because you use RAID is a problem.

 

[Steve_S]

"RAID is not a backup.

Disks break."

 

This is the whole point of RAID-1. If a disc breaks, you don't lose any data. You may be confusing RAID-1 (mirror) with RAID-0 (stripe) where two discs share the read/write duties for increased performance. If you lose one disc, you lose all the data, unless you've mirrored the array (RAID-0+1).

 

But we're just talking about your music collection and the effort you put forth to rip it, and the backup requirements are a lot less rigorous.

 

If you're talking about a hard drive that ONLY holds music then I agree, an external automotic or manual backup is the most appropriate. But you're failing to think about all the other data on the system including personal files, business content, email, etc. If I were to lose the last 3 hours of new data it would be a very big deal for me. RAID-1 is the best solution by far, and an external or remote backup protects against further risk.

 

This shouldn't be presented as a false choice between not doing backups

 

No one is suggesting anything of the sort. Quite the opposite in fact. RAID-1 is a real-time backup. External storage is a safeguard backup. Remote storage is also a safeguard backup.

 

 

[PeterSt]

"RAID is not a backup.

Disks break."

 

This is the whole point of RAID-1. If a disc breaks, you don't lose any data.

 

That's why I suggest to use both RAID-1 and a normal backup.

 

If you're talking about a hard drive that ONLY holds music then I agree

 

Uhm, anyone not doing it like that doesn't do it good anyway. Why ? because of your arguments.

 

WoodsWeller,

 

If you have 2000 discs ripped and you back them up, and you buy and rip 25 new discs over three months, and you lose your data and have to restore from a three-month-old backup, oh noes! you have to rip 25 discs over again.

 

Let's see. I started ripping maybe 5 years ago. That would be 60 months, devided by 3 = 20. That times 25 gives 500 CDs. Now something must be wrong with your assumptions, because I have 12,000. That's 24 times more than you thought. IOW, that is around 25 per day.

Yes, I have piles waiting, because I am behind. I don't want to get even more behind.

Oh, I can listen to 4, maybe 5 per day. So I'm definitely doing something wrong.

 

I'm only making fun of course, but I'm not kidding.

 

I guess we now are debating for the debate. Time for some listening !

Peter - Out

:-)

 

[Steve_S]

"If you're talking about a hard drive that ONLY holds music then I agree

Uhm, anyone not doing it like that doesn't do it good anyway. Why ? because of your arguments."

 

What? I have no idea what you're trying to say. The point of this topic is whether or not RAID can be used for data backup. The answer is yes. Whether or not it's the right type of backup for a particular system is arguable. As a backup for a data disc that is only updated every few days it is absolutely a waste of resources for most people. As a backup for an entire system it's very appropriate and invaluable in the event of drive failure. No localized backup solution, especially an integral real-time backup like RAID-1, should be relied on as a sole means of data security.

 

The number of CDs you have digitized is irrelevant to the type of backup that works best for your system.

 

[PeterSt]

I am sorry Steve, but I have the feeling that you turn everything exactly upside down. I can't see any logical responses to what I said. But then there's by english ...

 

You are right.

 

[Steve_S]

It's your English I'm having trouble with! If I've misunderstood what you're trying to say then I apologize, but the impression I get is that you've spent a bit of time contradicting yourself and then trying to back up when called on it. I don't mean to be argumentative, I just don't want misinformation floating about whether by typo or other means. My own posts are not immune to this belief.

 

If you are having trouble understanding what I've posted then I would be more than happy to elaborate, but first I would like to invite you to show an example of where I've turned anything upside-down! My points, in my view, have been only factual with no opinion interjected unless otherwise stated. If I have made any errors or was incorrect on any statements (discounting the typo you pointed out in my initial post) then I would appreciate correction. That's the way an online forum stays accurate and valuable as a resource!

 

[audioAl]

I have two full 320 gig Hdd's I wish to add to my music library, and I only have a 500 gig external drive for backup, guess I need to buy another?

 

[brucedecosta]

Not sure how much i appreciate your blog but i am great lover of your blog and visits it frequently! Do you think there are many more boards which are getting so much discussion just because they are posting interesting articles like you! company logo design|online logo designHope you keep this in future as well!