Would love to get some feedback/advice on Backup Strategy for Aperture. Long post, but this is a subject which should be of concern to all photographers….
I got a big slap in the face the other day when I opened my Aperture library and saw that EVERYTHING had a 5 star rating. As I am rigorous about backing up daily using CCC, I didn't skip a beat and opened my backup library.
Oh uh.. That one was messed up as well. Now my heart started beating faster, but I still had two more backups. But the next one was all 5 stars and the final backup on my Drobo also was all 5 stars. Now my heart is starting to race… really, really fast.
Fortunately I had made a new library a few days earlier and imported my existing library into it (to try to address a bug where Aperture is constantly regenerating thumbnails, but that is another issue.) I hadn't yet trashed the original library and to my enormous relief all my ratings were intact. I wound up losing two solid days of work, a tiny price to pay. But I did get a giant bucket of ice water poured over me and my backup strategy.
For now, let's leave aside the issue of “why did the library get to be all five starts?” (which I've never heard of Aperture doing all on its own so will have to take credit for human error. It's actually really easy, just select all your photos and press the “5” key. Poof, a year of hard editing work down the drain….)
I'm grateful for the event as it has forced me to completely rethink my strategy.
Here's my scenario:
- 2 distinct 100GB Aperture Libraries, one with my current project, one with everything else, each with about 2TB (approx 100,000) referenced images.
Old Strategy:
- Daily clones with CCC to two different backup RAID's and to a DroboPro
- Weekly backups to a pair of Offsite drives
- Rotating offline backup drives shipped every few to my mother's house in New Mexico, far from earthquake prone San Francisco Bay Area.
New Strategy Ideas:
- Continue the Daily clones, but with the Archive command in CCC or ChronoSync (I was using the Archive feature in the past, but for some reason had it turned off recently)
- Turn the DroboPro into a (slow) TimeMachine and let it run at night. Would take forever to restore, but better than a straight clone.
- Rotating Offsite drives - I've got 2 x 4TB drives “Set A” and “Set B”. One of them is always at home, away from my workplace. Each day I take one set and pop it into my RAID tower, update it and then bring it back home. This way, even if my office burns down, I've got a backup offsite.
But this is not enough as my home is still in the SF Bay Area and if disaster strikes, I could lose both home and office. Not likely, but I realized after my little 5 star incident that my entire life's work is on these drives and I can't be safe enough…
So I've been thinking about an even more robust strategy for Offsite backup and would love to get your thoughts:
Option 1 - Buy a Synology NAS with dual disk redundancy, copy all my data to it at my location, send it to my mother's home in New Mexico, and put it behind a SonicWall router with SSL VPN connection. Total cost would be about $2,500 for 8TB storage so along with my photographs, I could put ALL my files including my home folders, music, movies etc there. I could use ChronoSync to backup to the NAS from anywhere in the world, and would implement an archive strategy for Aperture so that I'd get incremental backups and could roll back my library to a given date.
Pro = Total control. Tons of space. Can also use as a fileserver for my business, giving access to others as need. In emergency situation, the entire unit could be FedEx'd back to me for restoring.
Con=Initial cost. Limited transfer speed due to consumer speed internet connection. Data in one location, not distributed among multiple data centers. Susceptible to occasional power outage, etc.
Option 2 - Use Backblaze/Crashplan -
Pro = ???
Con = Will take forever to seed. Concerned about data integrity after reading some articles online. Unknown ability to save multiple versions of an Aperture Library.
Option 3 - Use Amazon S3 with Arq client to backup Aperture Libraries - Send offline hard drives with referenced masters to New Mexico.
Pro = Amazon reliability. Ability to only backup changes to Aperture Library using Arq. Low cost as only backing up Libraries, not the masters. Can send seed drive to Amazon.
Con = ??
I did read one good strategy to backup multiple versions to an Cloud solution such as S3, which is to have ChronoSynch backup the changed files to a folder and then upload that folder to the cloud along with the updated library.
So lots of options… And one important question which seems to be missing from most discussions I have been reading about Aperture Backup, which is RESTORING Aperture Libraries! I've read plenty of horror stories about restoring from TimeMachine. But what about from these other methods? And how does restoring to a previous version of a Library work when using ChronoSync archive feature?
For example, lets look at my recent 5 star mishap. If I had been backing up with ChronoSynch with Archives, how would I restore to a previous version? Is it simply a matter of restoring the subset of archived files back to the Library, starting with the most recent, until my normal ratings are restored? If this is the case, seems like one could create their own reliable TimeMachine by running a ChronoSync Archive every hour or two throughout the day.
Would very much appreciate your thoughts and suggestions on this matter. I hope that this discussion is useful for others, as we are all
ultimately in the same boat entrusting our images to hard drives no matter where they reside. Thank you.
(PS. I'm going to cross post this over at Apple Discussions…)
The 5 stars on all images ‘misstep’ will not happen if you select the option to only edit 1 image at a time. (Rectangular button with a ‘1’ in it).
RAID 1, in my opinion, is a waste of time for photographers. It’s only ensuring against HD failure. It’s useful for banks etc., or anyone who cannot afford to have any down-time, but photographers don’t fit this scenario. RAID 1 isn’t a second backup or useful mirror, in the sense that any errors you make during your backup, are immediately replicated to the mirrored raid drive … as you discovered with the 5 stars. So you still need a second backup.
For us, we’re mostly sitting at our desks when any backup that matters is happening (i.e., something new to backup). Should a disk fail (happened to me once 15 years ago, in 25 years of computing) then you can instantly run a repair utility and if that doesn’t work, order a new disk and in the mean-time, switch to your 2nd backup … which is of course, sitting next to you on your desk, or like me, is attached via ethernet, in a fireproof safe, in the basement of an adjacent building.
There’s no magic bullet for backups and just like the 5 star misstep, it’s down to us what we backup and down to us not to mess up the backups.
Sending copies to your mum sounds good … so why can’t you get one of those back? Too far away and too slow I guess. This highlights the difference between backup and archive. They’re not the same thing and we need both. Cloud is great, if you have fast enough internet and patience like Joseph, but clearly it’s not yet feasible for many.
I use Vaults and don’t see any reason to want to replace that system. It’s simple and built-in to Aperture. I also archive Projects, simply exporting them as Libraries. This is a really useful thing to be able to do, because you can keep the whole job in one place if you have bits and pieces from different apps (as I do).
I’d attach two Vaults as HDs and simply press the ‘update vaults’ button at the end of each day, or perhaps twice a day. That way you can only loose, at most, one day’s work and you have redundancy because of 2 vaults. They’re fast and efficient. Of course, they don’t cater for Referenced files, so those you must backup those with another app., Time Machine is not ideal for these huge files/folders. Personally I prefer to work with individual Libraries, than Referenced images (too much danger of human error in the Finder) so I have everything Managed per job, which is then archived as a Library. In my main Library, I then remove all but the 4 and 5 stars from the job, once it’s over and delivered, in order to keep it small enough to have all my finished/delivered images and ‘near keepers’ in one, constantly available Library. I keep this on an internal drive on my portable, so that no matter where I am, I have all my delivered files with me.
Works for me.
Hi Grant,
Thanks for your sharing your strategy. I’ll consider using the vault as well.
Someone pointed out to me that my post was too long and without specific questions, so to be clear:
Questions (all subjective of course…)
-What is the best strategy for managing onsite backup of multiple +100GB Aperture Libraries, and +4TB of referenced masters that accounts for both multiple drive failures AND potential data corrupution?
-What is the best strategy for managaing offsite backups (NAS accessed over internet, Amazon S3, Backblaze/Crashplan).
-What is the best way to maintain a rolling backup of the Aperture Library in case of data corruption and best procedure for restoring. (Time Machine has proven less than reliable for this and would like to implement another system in addtion to TM.)
Thank you !
"There is nothing worse than a sharp photograph of a fuzzy concept." Ansel Adams