Backups, External Drives and RAID... please HELP!

You might want to get something like this 4 bay external hard drive enclosure and just add the hard drives yourself. If you look in the Frys ads, there are usually hard drives in there that are cheap.

 

Easiest answer - Forget RAID.

RAID is not backup. If you delete, overwrite or corrupt a file on your main drive, it is b0rked on the RAID mirror immediately as well.

RAID 3 or 5 (or 1+0) on an external multibay drive offers you some extra reliability in that it will survive the failure of one of the drives in the array... but it is expensive.

Get a nice large external Firewire drive to back up to, or one of the network attached storage drives. Then set up Superduper to manage your backup schedule.

 

First, you should do just a little bit of research to get a proper grasp on what your looking at.

Wikipedia gives a good, brief run down of what raids are: http://en.wikipedia.org/wiki/RAID

Check out raid levels. For non-technically minded, a "nut shell" description of the RAID types most commonly used in home/office type setups:

Raid 0: best performance when dealing with large files. Zero protection. One drive fails they all fail

Raid 1: Mirroring. This basically makes a copy of every file on one disk on another. It protects you from total data failure if a single drive fails.

Raid 5: Provides protection against single drive failure.

The advantage of RAID 5 over RAID 1 is the in-to-out disk space ratio. For example, in RAID 1 you can only use 50% of the disk space you put into the array. For RAID 5, you get approximately 2/3's or more drive space back. (e.g. in a 5 disk raid 5 you'd get 4/5's of the available space back)


Of note, RAID 5 is *not* RAID 0+1.

Finally, as mentioned before RAID arrays protect you against catastrophic disk failure or losing all your data because a disk dies. It does not protect you against "oh ****" situations, such as when you accidentally delete your project directory a special kind of backup called an 'incremental backup' protects against this. Also, neither a backup nor RAID provide a solution to the "hmm, I need the project files from 5 years ago". This problem is solved by a particular kind of backup called an archive.

The ideal solution protects against both these situations and may involve, for example, using backup software such as Retrospect or free solutions like BackupPC to back up to a RAID 5 array of disks.


All of this may seem overwhelming, but there are a few VERY simple and NECESSARY questions you must answer *before* you even think about the solution you need:

1. How much data do you have to back up?
2. By how much does this data grow at what pace? (increases by x gigabytes every y months)
3. How long do you want incrmental backups for? (Hint: Forever is definitely the wrong answer. )

Answering these questions provide the starting point for knowing what you really need for your backup system. For example, if it's taken you 10 years to build up 500gb and you only want immediate access to the last 6mths of work which amounts to 100gb of data (in case of accidental deletion). It may be gross overkill to build a 1.5 terabyte (1500 gigabyte) RAID 5 array. A more appropriate solution may be an off the shelf 1TB mirrored array (with 500gb usable) AND a robust archiving policy for the 400gb of 'old data' that you save just in case or for records purposes.


Just some food for thought. Start with the 3 questions, backups aren't a penis measuring contest so don't let folks convince you you need a Mercedes when a Corolla would do just fine. The best way to avoid that is spending an afternoon on wiki or some such reading about raids, backups, etc.

Oh, I've always thought the backuppc docs provide some good plain talk backup basics in their docs. You may want to peruse those sections: http://backuppc.sourceforge.net/faq/BackupPC.html#backup_basics

 

Get a 500GB external hard drive, or even better a 750GB one. You can find really cheap ones if you look, get with it a USB 2.0 or firewire enclosure which you can get for 30 bucks. It will be much cheaper then buying a named brand and you build it which will make it better.

 

That is a very modest setup. I would not waste the money on a RAID 5 (which requires minimum of 3 disks and enclosure, card, etc.)

I would go with a mirrored disk solution. Something off the shelf like the Onetouch III would work nicely. I picked one fo these up for a researcher recently. It comes with retrospect lite backup software or you can use it as an external disk with your own backup software (like the soon to be time machine)

(http://www.newegg.com/Product/Product.aspx?Item=N82E16822144071)

The Onetouch model above in mirrored mode would give you 500gb of useable storage or about 460gb after formating. This will provide enough for your total of 250gb plus a little room to grow, which is important.

I would use the onetouch to do incremental backups with a 30 day retention. This means you would be able to "go back 30 days" for accidentally lost or deleted files. A good backup program should be capable of doing 30 days of backups for 250gb on about 300-350gb of disk space. Though this number depends heavily on how many files you create or modify over that time and how well your files compress.

For the 100gb system drive, I'd just buy another 100gb drive which are incredibly cheap nowadays and create a mirror with your system drive. OSX is capable of doing this "out the box" without any additional hardware or software.

With that setup, a total disk failure on your computer would be no more of a hassle then rebooting. The only thing missing from the above is an archival policy...or "putting it on the shelf". Admittedly this is somewhat of a hassle for today's home user. For example, the initial backup of your 150gb only would create a respectable tower of dvd's.

This is what I would do myself:

1) Monitor the incremental backups on the 500gb (1tb mirrored) drive. Is the total size at the end of the month relatively the same as the first day?

--) If so, perhaps a 1 year "archive/retention" is enough and you can simply increase the time to expire (these are all backup settings in most software)

--) If not, keep the retention low and invest in an archive system. In this case, I'd purchase a single drive capable of *just* holding a full backup every 6mths or so and dump a full backup onto it. Then label it appropriately (date, data, etc.) and put it on a shelf in an electronically friendly place.

Remember, none of these are required and are matter of how much your data is worth to you as far as the extent of money and effort to expend. At home, I don't do archives. The data, realistically, isn't that important on the whole. I burn things like financial records, tax returns, etc. to dvd since they take very little space. . . but that's it.

You may find that the onetouch and system disk being mirrored take care of all your essential needs and the archive is too much hassle/money to bother with.

 

I have a Raid 5 (4 x 250gb = ~700GB usable space) setup on my pc using the XP hack described on tomshardware. That has worked fine for me until my PC started acting up....I think something (other than the harddrives) is about to die and since I'm not ready to get a new computer I bought a 1TB hitachi drive (usb2.0) and backed everything up to that. That set me back about $400 at bestbuy. When I do get a new desktop I plan on continuing to use the raid 5 as my everyday storage and have the 1TB drive as it's archive.

 

I'm not sure if your replying to my post, but the set up I described would prevent any data loss unless up to two drives physically fail at the same time.

Further, it prevents down time. By mirroring your boot drive in a RAID 1, your system never goes down for more then the time it takes to reboot.

You don't want to get a 160gb most likely. You want a 100gb, or same size, to mirror your boot disk. Otherwise you will either a) lose 60gb of space or b) have an extra 60gb partition just hanging out if you use apple's software raid and mirror against a partiition.

I suggested something like the 500gb mirrored for a simple reason:

It takes more space to back up then what you have to actually back up, especially if your files change a lot.

For example, with the 250gb drives in a mirrored configuration you'd only be able to use about 225gb. If your a graphic designer and are constantly changing files and creating new ones you could easily use all of that with a simple two week backup policy. Then your looking for new drives and expanding space 2 weeks after you built your backup system.

If you want to go the economy route where you buy the bare minimum extra parts do this:

-) 2x 250GB drives in a RAID 1
-) Buy that 160gb drive and put that into a RAID 1 with your main hard drive. This should be inside your machine if at all possible.
-) Use backup software. Simply copying or manually backing up is tedious, inefficient use of disk space, and does not protect you against things like accidental deletion.