Backup Strategy

I’m paranoid about backups and review my overall backup strategy at least once a year. It’s that time of year again. While it’s geared to those like me, home and small business users with a few computers and mobile devices, the overall philosophy is suitable for any one or any size business. It’s just the solutions that will vary.

My Backup Philosophy

  1. A file doesn’t exist unless it’s in three places.
  2. At least one of those places must be geographically distant from the others.
  3. RAID, or a RAID like device, such as a Drobo only counts as one copy. Even if that one copy is redundant any corruption or deletion removes the file. It’s also possible for the device (not the disk) to fail.
  4. The backup has to be automatic and unobtrusive. If I have to remember to run it, it won’t be run.

While data storage is getting cheaper all the time, it can still be expensive or infeasible to store terabytes of data. So I classify my types of data.
Critical: Financial information that would be required to recover from a disaster. This would also include data for current business projects. The idea here is that in the event of a disaster (as opposed to a simple PC failure) I have the information needed to recover as fast as possible. This is the data I need without delay.
Important: Data I don’t want to lose but I can wait a day or two to get it back. This is data such as archived financial records, business records, family photos and family movies among other things.
Pack Rat: This is data I save because I’m a pack rat. I probably won’t miss it if it’s gone. While I do want backups, if the data is large I may decide the data is not worth the hassle or expense of offsite storage. This includes movies I’ve ripped from my DVD collection and some really old personal and work files.

I also want to be sure my computer can be up and running as soon as possible after a hardware failure. Along these lines, getting the data to new hardware quickly is a priority. Some things to consider for this are:
Bootable Backups: A local backup to an external drive that can be immediately plugged into my Mac. I should be able to boot from that external drive and run everything until I have time to replace the bad disk in my computer.
Alternate Hardware: If your computer is critical to you business you’ll want to have a spare quickly available. It doesn’t have to be a complete replacement, but it should be able to keep you going. I have a desktop and laptop so I make sure I can access all data from both computers. My iPad can also fill in for many of my computing needs in a pinch since much of what I do is web based or plain writing.
Portable Data Formats: I try to keep my data as portable as possible. I’ve been using Markdown so that my writing can be in plain text, while still being formatted, allowing me to use any computer (OS X, Windows, iOS, Android) if needed. I may not have my apps of choice, but I could still work. Web based email using IMAP also helps. I can access email anywhere and it updates everywhere else. For critical data, I want to avoid being locked into a program if at all possible.

Some of this may not be financially feasible, depending on what you need. If you do intense CAD work or movie editing having a spare machine may not be the best use of your money. It may be better to bank the money and have a plan to replace any bad hardware as soon as possible and bite the bullet to deal with any productivity interruption.

My Strategy

  1. I centralize my file storage, using a Synology NAS (Network Attached Storage) to store my Critical and Important files. This makes it easy to find these files and keep them backed up. This doesn’t require a lot of storage and is currently just 14 GB of Critical data. Critical data is actually less than the 14 GB since I could wait for a restore for most of it. But the total is relatively small and separating it out would complicate my workflows. I also consider photos and music as Important data which adds another 130 GB or so. As video becomes a bigger source of critical (business) data I do expect my Critical data needs to grow.
  2. I use ChronoSync (website) to synchronize my most Critical data from my Synology to my Mac desktop and Mac laptop (when it’s online and on the local network) every night. Changes are infrequent so this sync only takes a few minutes and it makes sure I have a second (and third) copy of my files ready to use if my Synology NAS goes down or I lose my local area network. This sync also archives any files replaced due to a change or deletion. I have this set to save deleted/changed files for 180 days although I may have to lower this if disk space becomes an issue.
  3. I use Arq Backup (website) to backup the critical files every hour to Amazon S3. Again, since changes are small the time needed is minimal so I don’t even notice the backup. Arq keeps copies of all updated and deleted files so I can get a file from the past. Files aren’t deleted until Arq runs out of the space I’ve told it to use. I let Arq use enough space to keep about a year of my critical files.
  4. I use Synology Cloud Station, a Dropbox like service but one that’s totally under my control since it runs on my Synology NAS. I use this to save work in progress and other files I want handy. It syncs automatically to all both my computers and my NAS. The Cloud Station files are both Critical and Important so I treat them all like Critical files to make management easier and changes are backed up to Amazon S3 every hour. Cloud Station will also save older version of files, along with deleted file.
  5. All my Pack Rat files, except my videos, get backed up to Amazon S3 once a day using Arq Backup. I do have a small number of work and personal videos I’d like to save so they are included in the Pack Rat files.
  6. My videos are rips of the DVDs I own to make them easier to view. With the growth of streaming this collection really isn’t growing anymore and I’m not willing to jump through hoops and spend any money to protect it. But I did copy the videos to some old hard drives and I store one in my house and a second at another location.  If my home is destroyed I’ll have bigger concerns but could eventually get the videos back.
  7. My MacMini does a Time Machine backup to my Synology NAS. It also does a nightly Carbon Copy Cloner disk image backup to a external drive that’s bootable. Every two weeks I do another Carbon Copy Cloner image that’s stored in my truck. (Yes, that’s my local offsite backup.) I’d rather have it close, just not near the computer and if I’m away when disaster strikes my car is probably with me. I used to keep it in an office or a friends home but then access is a potential problem. Plus keeping it updated was a hassle so it became outdated. Even though the backup is up to a week old, it’s bootable and then the newer data is available from my cloud storage. The drive is encrypted so if it was stolen it couldn’t be used. Plus, nearly all my data is on my Synology NAS so this drive is only needed to quickly get the Mac going again.
  8. My laptop gets a Time Machine backup to my Synology NAS when it’s on at home.Every two weeks it gets a Carbon Copy Cloner disk image backup that’s also kept in my trunk. While I do have to remember this, breaking rule #4 of my backup philosophy, I automated as much as possible so all I have to do is wake the laptop and plug in the external drive.
  9. I schedule quarterly tests restores of each backup and also make sure the bootable backups do in fact boot. To be honest, a quarterly schedule is one thing, actually doing them is another and I do it about twice a year.

Items 1, 2 and 4 in my backup strategy are designed to make sure I can pick right up on my second computer should one of them not be available.

Naturally, if I’m traveling with my laptop and it fails I have a problem. Even if I had one, carrying a second laptop wouldn’t make any sense. But I’ll have to consider some strategies for dealing with some of the more common failures such as a hard drive.

Overall I’m confident that my strategy can protect me from losing my data. Replacing lost hardware in the event of a disaster is a different problem, although no less important. I still don’t have a quick solution to that but at least I’ll have the data.

My strategy works for me. While I think the philosophy is immutable, how it gets implemented can vary by user. Due to the file syncing my solution is a little complicated. But it works for me. You may find a different solution works for you.