Need backup advice?

[Zyniker]

Zyniker,

Backup: A copy of a program or file that is stored separately from the original.

RAID doesn't do that.

Read this article:

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

It doesn't mention RAID at all. Why? Because RAID is not a "backup system" or "backup plan". Someone thinking that RAID 1 offers them a backup system or paln doesn't make it a backup system or plan.

S-

*sigh*
RAID 1 copies the data from one drive to another (hence that whole "mirroring" thing). Since you like Wikipedia so much:

http://en.wikipedia.org/wiki/RAID_1#RAID_1
"A RAID 1 creates an exact copy (or mirror) of a set of data on two or more disks. This is useful when read performance or reliability are more important than data storage capacity. Such an array can only be as big as the smallest member disk. A classic RAID 1 mirrored pair contains two disks (see diagram), which increases reliability geometrically over a single disk. Since each member contains a complete copy of the data, and can be addressed independently, ordinary wear-and-tear reliability is raised by the power of the number of self-contained copies." (emphasis mine)

Last I checked, each hard drive in a modern computer is physically a separate device from other hard drives. So, what part of your little definition doesn't RAID 1 fit? It's a copy? check. It's separate? check.

Please, just because you seem to believe that RAID 1 isn't an appropriate backup doesn't mean it isn't a back. As we've agreed, it's a bad backup plan/system; however, it is still a backup. Yes, it's located in the same computer; yes, it's more prone to failure than a correctly designed backup system; and, yes, it is simply a mirror and not an archive (which isn't technically a requirement to be a backup, it's just recommended and sensible). Nevertheless, RAID 1 is a backup regardless of "omeone thinking" it isn't.

 

[pprior]

You guys are sounding like Bill Clinton arguing about what the definition of "is" is....

Clearly Raid1 is a backup. It copies data to a separate device/disk.
Clearly it's not sufficient as a reliable backup for many reasons outlined.

It would provide data backup if one of the disks failed, it would not provide backup if the data got corrupted or deleted.

Let's not get caught up in semantics, just state the practical facts and let people decide for themselves what is best.

The real question you need to answer when arguing this question is this:
What proportion of data restoration is caused by:
1) Drive failure
2) System failure (power surge, etc)
3) User error
4) File corruption
5) Need to revert to older version of data

If you can answer that question, then you can decide whether RAID1 is a -sufficient- backup or not. I think most serious folks would say it's clearly not.

 

[Mhaddy]

I am struggling with two backup issues. First space. I have 4 TB growing to 5 TB of digital media, %95 photography. At a minimum I need two copies, plus the original.

...

Second question what about back up software? Time Machine is interesting but will only create first back, not the second. We like it as it keeps the files in native state so we can easily find them if we need to.
We tried Super Duper and it had a few bugs. Does Chronosync work in large data sets? I see that EMC is about to ship a new version of retrospect, but I hate it when these back up systems create their own data structures.

thanks for you help in advance.

I hope you're using a good photo management software like Aperture or Lightroom to store your photos as it will maintain your original and whatever versions you create. I then use its built-in utiliy to automatically sync & maintain a duplicate archive on my external drive. So I don't use ChronoSync for this, but I do use it to b/u my iTunes library, PSD's/AI's and movies -- and it works splendidly.

 

[rylin]

You guys are sounding like Bill Clinton arguing about what the definition of "is" is....

Clearly Raid1 is a backup. It copies data to a separate device/disk.

I hope you enjoy wiping your ass with cash.

 

[Sesshi]

I hope you enjoy wiping your ass with cash.

Or blubbering "But the computer deleted it!"

 

[fatespawn]

First Post too and I'll throw in my 2 cents.

I'd recommend trying to keep everything on one of your 1TB drives (if you have the spare room). Here's how I would set up your system - opinions may vary.

First clone your 500GB boot drive with SuperDuper! (or one of the other recommendations) to one of the 1TB drives and set that one as the new boot drive. Then copy your data from the other 1TB drive to the new 1TB boot drive so all data is on the boot drive.

now, wipe your 2 500 GB drives and stripe them to make another 1TB drive.

Set Time Machine to use the other 1TB drive as it's target
Use SuperDuper! on a schedule, to clone your boot drive to your new Striped 1GB drive. You can set it to clone nightly and shut down your computer when finished.

The stripe is twice as likely to fail, but no big deal since you'll still have your original boot drive and you can just swap out both 500's in the future if you want. If one of your 500's fail, you'll lose your time machine backups. If your boot drive fails, you just boot from the stripe and slap in a new drive and clone back to the new single drive.

This method would give you incremental backups through time machine and would give you a cloned backup to get instantly up and running in the event of a boot drive failure.

In most scenarios, the only thing that never gets' backed up is time machine.

Comments?

-fate

 

[sidewinder]

RAID 1 copies the data from one drive to another (hence that whole "mirroring" thing).

Actually, RAID 1 writes the same data to both drives to provide high availability through redundancy in case of a drive failure. If one of the drives fails and the array is rebuilt, then the data is copied from one drive to another.

Anyway, you can keep saying that RAID 1 provides a backup if you want. But that does not make it so. Yes, a backup is intended to protect someone from a drive failure. But a backup does much more than that.

S-

 

[nanofrog]

As sidewinder mentioned, RAID 1 is meant for increased uptime (availability). The computer is still up and running if a drive fails.

Backups, (archival storage), is part of a disaster recovery plan. This is meant to deal with recovery when major problems happen. And they will, unfortunately. No system is 100% fail proof.

 

[NoNameBrand]

Those of us who actually value copies of our data don't use RAID1 as backup.

Since no one asked, here are some backup schemes I use personally and professionally:

Home workstation:
SuperDuper! clone of every volume that isn't scratch or itself a backup (4/13 total volumes cloned). Runs nightly.
TimeMachine of user files (excludes apps and media taken from my on-disc collection), runs hourly.
Mozy backup of the same dataset as TimeMachine. S3/JungleDisk/et al. would work too, runs nightly.

Personal laptop:
TimeMachine of userland data to workstation
SuperDuper! of disk to workstation.

Work Mac:
TimeMachine to a Linux box running afp.

I use rsync to backup a couple of Linux servers at work, hourly. Purportedly, rsync can now work quite well backing up Mac OS X to Linux with the latest versions of rsync fully supporting resource forks and whatnot.

 

[pprior]

Or blubbering "But the computer deleted it!"

Did either of you rude people even read my message?

 

[rylin]

Did either of you rude people even read my message?

Yes we did; especially the part where you reiterated that RAID-1 is a backup.

 

[rylin]

Mozy backup of the same dataset as TimeMachine. S3/JungleDisk/et al. would work too, runs nightly.

FWIW, I find Backblaze is a lot faster for me.
Mozy is capped (by the application) at 1Mbps but gives me around 500Kbps.
Backblaze is uncapped, and I get around 500KBps.

Still nowhere near my maximum (I generally hit 13-15Mbps upstream), but acceptable for $5/month.

 

[pprior]

Yes we did; especially the part where you reiterated that RAID-1 is a backup.

Sigh...apparently you didn't read the last part of my message where I agree with what your ultimate point is. Your brain must have been too busy thinking up your next insult to process it.

Substance people, SUBSTANCE. Get to the meat.

 

[rylin]

Sigh...apparently you didn't read the last part of my message where I agree with what your ultimate point is. Your brain must have been too busy thinking up your next insult to process it.

Substance people, SUBSTANCE. Get to the meat.

You saying RAID-1 is a backup doesn't make it so.
That's the part I take offense at.

As for the insults, I see more from you than from anyone else, Pot.

What's worse is that in a thread like this where the OP doesn't quite know what RAID actually is, people are saying RAID-1 is a backup.
That's just pure insanity.

Here's a hint: it can be a good idea to put your backups on a RAID, but no matter how much any of you try to spin it, a RAID is not a substitute for a backup.

 

[sidewinder]

Sigh...apparently you didn't read the last part of my message where I agree with what your ultimate point is. Your brain must have been too busy thinking up your next insult to process it.

Substance people, SUBSTANCE. Get to the meat.

You waffle like a politician! Earlier in this thread you wrote:

"RAID1 will save you from a single drive failure, and that is a good thing, but it is NOT backup."

Then you wrote:

"Clearly Raid1 is a backup."

Well, which is it??

S-

 

[NoNameBrand]

FWIW, I find Backblaze is a lot faster for me.
Mozy is capped (by the application) at 1Mbps but gives me around 500Kbps.
Still nowhere near my maximum (I generally hit 13-15Mbps upstream)

My upload speed is 1.5Mb/s. (and 15Mb/s down), so Mozy is just fine. I signed up for Mozy before there were many alternatives on the Mac; I may look around and see what else is available when my Mozy is coming up for renewal later this year, though, or when I buy a new Mac (we'll see if my account transfers).

 

[pprior]

You waffle like a politician! Earlier in this thread you wrote:

"RAID1 will save you from a single drive failure, and that is a good thing, but it is NOT backup."

Then you wrote:

"Clearly Raid1 is a backup."

Well, which is it??

S-

Wow you guys really have a lot of emotional energy into this discussion - either that or you need a better night sleep tonight than last night.

I don't think you've seen any insults from me, so not sure what the above posting was referring to.

I'll say it again, as CLEARLY and SLOOOWWWWWWLY as I can... turn off the insult center of the brain and please read:

1) Making a duplicate copy of any information is a backup
2) Having said dupicate copy immediately tied to the first disk is a very limited backup, which ONLY protects against a single drive failure and not the other whole host of issues that more frequently cause data loss.
3) Thus while Raid-1 -IS- a backup in the literal sense, it is not a logical backup STRATEGY in the functional sense, and is not one that I use for those reasons.

Personally I have my critical data backed up on a drobo time machine volume, separately backed up on a separate raid5 NAS in the basement, on rotating hard drives stored in a media fireproof safe stored within a larger fireproof safe, and offsite at mozy. I think that ought to make clear to you my idea of whether raid-1 is a smart overall backup strategy.

I don't know how much clearer I can be for you and I question why everyone's panties are so tight on this topic. I'm not selling anything, and other people can do what they want. I applaud those of you trying to get the word out that raid-1 is not a viable backup strategy, but I really question your means, which turn a lot of people off from reading your message.

Edit: I went back and edited my initial post that you reference to make my position more clear. We may disagree but I don't think I can be more clear than I have.

 

[rylin]

Wow you guys really have a lot of emotional energy into this discussion - either that or you need a better night sleep tonight than last night.

Personally, I tried detoxing to quit my oxycontin use last night, so I guess there's something we agree on, at least

As for making a duplicate copy; no, that's not a backup.
If I attach a photo to an email and send it off to someone else in my family, it's not a backup, despite being an exact copy.

Why?
Because it's not done in place of a backup, but it would seem you'd have it the other way.

In the same way, mirroring is not a backup, despite being a 1:1 copy.
This is the same situation I see with a lot of the people in my industry -- they think having a fully replicated DB server is a backup.
Guess what? It's not!
The second you shoot yourself in the left foot, you get a glaring hole in the right foot as well.

You can call it arguing semantics all you want, but it doesn't change the facts:
RAID is not backup.

 

[pprior]

Ok despite swearing I wouldn't argue with you anymore, I just can't quite understand what you're trying to say, so I'm going to come back ONE more time

Please explain to me WHY sending the photo via email is not backup. Now for the sake of discussion we are going to assume that the recipient is not going to delete said photo, correct?

Step 1) PHoto is emailed
Step 2) local copy of photo is lost, for whatever reason
Step 3) ask for photo to be emailed back
Step 4) Photo is back, and data is restored

Please explain why that is not backup? I don't care about dictionary definitions, what I care about is having my data when I need it, and that is what everyone else here cares about as well.

 

[rylin]

Ok despite swearing I wouldn't argue with you anymore, I just can't quite understand what you're trying to say, so I'm going to come back ONE more time

Please explain to me WHY sending the photo via email is not backup. Now for the sake of discussion we are going to assume that the recipient is not going to delete said photo, correct?

*snipped list*

Please explain why that is not backup? I don't care about dictionary definitions, what I care about is having my data when I need it, and that is what everyone else here cares about as well.

Thing is, you say you don't care about the dictionary definition -- which is quite obvious, since it would mean anyone who's ever published something somehow, somewhere has a "backup".

As for the email example; do you actually know that the recipient keeps all email? I.e., do you know the "backup" is still OK? Are you sure it's there when you need it?

A backup is worthless unless you need it, and if it's not available, it's useless.
If, on the other hand, you specifically email something in backup purposes, to a verifiably good service, then you might call it a backup.
Have fun trying to restore it though...

Mirroring, RAID-5, RAID-10 etc. is simply not a backup, no matter how much you want to spin it, as that's not its purpose. Not in any way.
The way you're putting things, you make it sound as if a glass of water is a fire extinguisher.

Again: RAID is for redundancy and performance. Go ahead and store your backup sets on a RAID -- that's a perfectly fine use for a RAID.

 

[pprior]

Ah, now YOU are backtracking and making assumptions about what another recipient may or may not due. Your statement was that email of the photo by definition was not backup.

This thread has run it's course. I think you're far too full of yourself when it comes to assigning what is backup and is not. If I email to my son and he archives his email on a local backup disk, then that IS backup. It may not be ideal, it may not be what the best case practice should be, but it is backup whether you want it to be or not.

Your definitions seems to be unless it's multiversioned, redundant, and physically separated it's not backup. While i agree with those goals, they are not absolute and not required.

Remember, the enemy of good, is better.

 

[rylin]

Ah, now YOU are backtracking and making assumptions about what another recipient may or may not due. Your statement was that email of the photo by definition was not backup.

This thread has run it's course. I think you're far too full of yourself when it comes to assigning what is backup and is not. If I email to my son and he archives his email on a local backup disk, then that IS backup. It may not be ideal, it may not be what the best case practice should be, but it is backup whether you want it to be or not.

Your definitions seems to be unless it's multiversioned, redundant, and physically separated it's not backup. While i agree with those goals, they are not absolute and not required.

Remember, the enemy of good, is better.

If your son backups his stuff, then yes, his stuff is backed up.
I'm not "full of myself", but I do take backups seriously (since, you know, it's part of what I do for a living).

Multiversioning, redundancy on the storage layer aren't strictly necessary in a backup (though, as you say, are good, and they tie in to the points that definitely matter).
Some measure of physical separation is definitely required, but differs depending on use.
My mom, for instance, would do just fine with TM backing up to a small external disk.

Me? Let's just say that the stuff I have backed up is too valuable.
I "need" a local backup and a remote backup.

The most important factors are availability and reliability.
If I can't initiate a restore at 4am in the morning, it's not a backup.
Do you REALLY want to call your son up at 4am and say "Hey, I need to get ahold of some of the files I've got backed up"?
A part of the whole availability aspect is indexing, so being able to find what you're looking for before restoring is a good thing.

The biggest point re: backups is it needs to protect against user error.
Protecting against a disk failure can be done with a RAID, but that's still just redundancy, not a backup.

Argue all you like, but I know what I get paid to do -- part of it is not giving stupid advice to someone asking about backups.

 

[Horst]

Ah, now YOU are backtracking and making assumptions about what another recipient may or may not due. Your statement was that email of the photo by definition was not backup.

Get a grip, dude , you are wrong, face it.
Happens to me all the time, take it like a man .

 

[pprior]

If your son backups his stuff, then yes, his stuff is backed up.

....

The most important factors are availability and reliability.
If I can't initiate a restore at 4am in the morning, it's not a backup.

...

The biggest point re: backups is it needs to protect against user error.
Protecting against a disk failure can be done with a RAID, but that's still just redundancy, not a backup.

....

No, rapid availability might be important to you and your clients, but it may not at all be critical for everyone. For many people, a day or more getting things back up would not be a critical problem, and may be a very acceptable trade off to save significant cost. For a business approach, uptime is critical, but for a home user, it may not be as much.

We all have different needs. For example: I use Mozy for part of my backup strategy. If I ever had to download all my data it would take DAYS to do. For your client that might be completely unacceptable and by your "4 am" argument I'm doing something wrong. But for me, the cost savings makes it an acceptable tradeoff to rotating out tapes to an offsite storage that I have quicker access to. One size does not fit all and defining everything as either "the way" or nothing doesn't help many users.

I agree with your last point as it relates to importance, and is the main reason I think time machine is such a good approach for the "masses" (including myself) who struggle to keep versioned backups of data.

And Horst... Never surrender man, never surrender

The funny thing is I would never recommend raid1 as a backup strategy, as I outlined earlier, but I find myself drawn to defend the semantics of it. I guess maybe -I- need more sleep.....

 

[rylin]

No, rapid availability might be important to you and your clients, but it may not at all be critical for everyone. For many people, a day or more getting things back up would not be a critical problem, and may be a very acceptable trade off to save significant cost. For a business approach, uptime is critical, but for a home user, it may not be as much.

Note that I said initiate a restore.
Even the fastest tape library would take a while to seek and unpack the right file(s) -- same with Mozy et. al. (and as you may have noticed, I mentioned using Backblaze myself).

We all have different needs. For example: I use Mozy for part of my backup strategy. If I ever had to download all my data it would take DAYS to do. For your client that might be completely unacceptable and by your "4 am" argument I'm doing something wrong. But for me, the cost savings makes it an acceptable tradeoff to rotating out tapes to an offsite storage that I have quicker access to. One size does not fit all and defining everything as either "the way" or nothing doesn't help many users.

To put things in perspective, I'm used to getting the most out of my 100/10 connection, so "days" is a strange concept to me. Hell, for $1500 - $2000 I get a dedicated gigabit pipe (and I've done so for a few clients), but when it comes to mozy, you have other options.

"Never underestimate the bandwidth of a station wagon full of tapes."

Mozy will gladly send you a HD or two with your restore on it. They'll probably courier it to you, so that project you were working on may well arrive before your 1pm presentation/whatever.

I agree that most of us can wait a bit with getting our restores, but it's a question of how long as well.
Is a year acceptable? Half a year? A month? Two weeks? A couple of days?
It all depends. For pictures of your dog/GF/family, you can probably wait a few days.. possibly weeks, but after that?
Availability is important because eventually, data gets stale.

Lets say you're working on a iPhoto book for someone's birthday, and you need that photo back from your son in order to order the book in time.
What do you do when your son is away on vacation?

In other words, while split-second availability would be awesome for everyone, guaranteed availability is much much more important.

I guess, to summarize I'd say:
There's no one-size-fits-all for backups, but there are key points that don't really change. So long as the key points are met, you have a good or great backup and recovery plan.

If you don't meet any of the points, or if any of the points are compromised, you don't have a backup or a plan