My Backup Plan
My hands-off strategy for continuous, redundant, backups
This week we had World Backup Day to raise awareness of the importance of regular backups. We all should be making regular backups of our data but, according to the statistics, very few of us backup at all, and the few that do, don’t do it on a regular basis.
Now, more than ever, we rely on —and generate tons of— digital data every day: photos, email, homework, documents, music, videos, etc., and all that data is stored in a computer’s hard drive. But imagine if said computer was lost, or the hard drive decided to call it quits. What would happen to that data? Will it be lost forever? That’s when you wish you had a good backup to come to the rescue. I don’t want to find myself in that situation, so, here is what I’m doing to prevent that from happening.
When planning my backup strategy I wanted it to be:
- Unobtrusive: I shouldn’t even notice that a backup job is running.
- Automatic: It should start without having to be manually initiated.
- Easy to restore: The restore process should be easy to use.
- Centralized: The backups should be stored in a central location.
- Redundant: I wanted to have a second, off-site, backup.
I wanted it to be this way because I wanted to keep it really simple. If I needed to connect an external drive to my computer in order to start a backup, I might do it every week, at best. Eventually I would forget about it, or misplace the backup drive, and go months without backing up.
All the backups needed to be in one location. I didn’t want to keep a stack of drives (one per machine), and it needed to be redundant with an off-site backup. Consider the case where the computer and the backup are lost at the same time — theft, flooding, fire, etc.—, if I had a copy of the data in a different physical location, I would still be able to recover it.
We have two computers (clients) and one server at home. All three are Apple computers running OS X. Time Machine comes as part of the OS, and since it’s easy to use and setup, that was the starting point of the plan. Time Machine saves hourly backups to an external or networked drive. Good.
The server (an old white 2008 MacBook that’s still going strong) is running Lion Server. As part of the server utilities, an external USB drive can be connected and made available for Time Machine backups for clients over the network. Perfect! After connecting and sharing the drive, I configured each of the clients to use it for backups and turned Time Machine on. That was easy! —So far, I’ve already met four of the plan requirements: unobtrusive, automatic, easy to restore and centralized.
Now, for redundancy I use a combination of Portable Home Directory (PHD)1 and Backblaze. I setup the server to make a location on an external drive available for Home Directories and then configured the clients to synchronize with this location. That meant having a copy of the client’s Home Directory data on an external drive in the server. Next step was to install the Backblaze client on the server and have it backup the contents of the server’s drive and the external drive containing the copy of the Home Directories.
How It Works Together
This is how it goes down:
- The client performs an hourly backup via Time Machine to the server.
- Every 15 min or so, the PHD sync job on the client kicks in; making a copy of the user’s Home Directory on an external drive attached to the server.
- The Backblaze client runs continually on the server, making a backup of the Home Directories and other data in the external drive to the cloud.
When considering an off-site backup service, it came down to two contenders: Backblaze and Carbonite. In the end I went with Backblaze because of its price. If I need to recover all of my data from Backblaze I have the option of downloading it for free —several gigabytes— or have a hard drive shipped with my data in it, at a cost. I hope it will never have to come down to it.
Please support this site, if you want to try Backblaze, use the link below. It will give both, you and me, a free month of service.
PHD keeps a user’s Home Directory synchronized between a network share and the local computer. If you are not on the network, you work on the local Home Directory. When you login on the network, the mirror agent running on the local computer synchronizes the two directories. ↩