Backups

There is a discussion going on in the Atlanta Flickr Group about various ways to backup your photo library.

I wrote the following, which after I wrote it realized it would be useful to post here as well.

I'll give you the rundown of what I do, but here's my basic advice:

1. Backup whatever you have to an external media regularly (i.e. daily or weekly)
2. Keep an off-site backup of that, and update often (i.e. monthly)
3. Get a real photo management package with database capabilities (there are many good ones)

I have 12 years worth of digital photos (history) and wouldn't want to lose it. Nobody does.

I recently switched from Windows (where I used ACDsee) to Mac and now use Aperture. Here's my basic setup:

Organization

• High level folder for each year
• Project for each day of the year (A "project" in Aperture terms is where the actual photo files live)
• Albums within each project (if necessary) to organize by topic, location, etc. (An "album" in Aperture terms is a container which holds references to the actual photos. A photo can be assigned to as many albums as you like)
• I can then create additional albums elsewhere to group and organize photos as I see fit, but never have to alter the actual physical location of the files (which is always chronological) because of the album tool/metaphor.

If you're interested, you can see a screen shot of what this hierarchy looks like in the Aperture Library.

Storage

• I keep everything on an internal 500gb SATA disk because it's fast.
• I don't run RAID, I prefer to use the extra disk to keep a seperate full copy (I'll explain why later)
• I bought a couple of external Western Digital MyBook 500gb Dual Interface drives at Costco for $229 each. Dual-interface because single interface only has USB 2.0, and IEEE 1394 (FireWire) is a hell of a lot faster (even at 400 mb/s) than USB 2.0 in actual performance. Though this may not make sense for you if you have a PC instead of a Mac.
• One drive stays at home connected to the system, the other sits in the vault at the bank. I rotate them monthly.
• I run two kinds of backups. 1) daily incremental file-level backups every night. Any file that gets added/modified will get backed up during the wee hours of the next morning. 2) application level backups in Aperture. Aperture has this concept of a "Vault" which is basically a full backup of your entire library (including database, modifications, etc.) that it can update.
• I keep one Aperture Vault on the other 500gb disk in my system, a 2nd one on the external WD MyBook disk, and a 3rd on the (disconnected) disk in the vault.

Benefits

You don't need to have a Mac or Aperture for this setup, though they do automate some of process.

Having the two-external disk method combined with the automated backups gives me the following advantages:

• Multiple copies of data. If I accidentally delete something tomorrow, I'll have today's copy.
• Incremental backups allow me to get something that was deleted 10 days ago, as long as it's within my retention window.
• Off-site backups allow for disaster (i.e. fire, theft) recovery (I keep my Quicken data on there, too) and also provide a longer retention window, in case you realize you deleted something last month that you didn't mean to.

RAID

I don't run RAID for a number of reasons:

• It adds complexity (complexity is the enemy of reliability)
• You have to test whatever RAID (1 or 5) setup you have to make sure that it actually works during a disk failure and that you can recover your data.
• RAID 5 (striping with parity) can slow you down
• RAID 1 (mirroring) doesn't have the performance issue, but you double your cost of storage.
• I decided to spend the money on an external disk and keep a full backup externally and off-site. This doesn't protect me as well in small windows of failure (i.e. between backup intervals), but it does offer better disaster recovery. I prefer this as my short term tolerance for data loss is fairly high (i.e. I could maybe recover today's photos from the memory card still, etc.)

Disadvantages

There are of course disadvantages to this approach.
• Cost may be one of them (i.e. you already have two internal disks you can use to mirror)
• Discipline. If you don't run the backups, you're not protected.

Alternatives

Finally, there are hybrid approaches too... For example, I also bought a Western Digital MyBook 1TB Platinum Tripple Interface (USB 2.0, FireWire/400 and FireWire/800) unit to hold video. This 1 Terrabyte of storage is actually two 500gb disks which you also have the choice to configure as a single 500gb mirrored (RAID 1) volume.

So in this case, you could have an external storage device that is mirrored. This one was about $449 on buy.com. However, I really also wanted off-site backups and felt I was protected well enough that I didn't want to spend another $449 to do that.