by Mike Shea on 25 March 2007
In my constant struggle to see how far my obsessive backup behavior can go, I decided to take another look at how I back up my data in my new MacBook inclusive world.
I had started simply enough. I have an external firewire disk hooked up called "Jonas", named after a hard caliber bad guy in the Dark Tower novels. Using Chronosync, an excellent Mac backup application, I am able to backup a current copy of all of the data in my home directory including documents, music, movies, photos, itunes downloads, library system files. Chronosync has a few things I really like including the following:
Chronosync does an incremental backup based on filesize and date. It only copies over files that have changed since the last backup.
Chronosync copies any data deleted on the source to an "_archive" directory on the backup. This means I can recover files I deleted a month ago but I also have a pure copy of my current home directory if I just want to restore current stuff.
Chronosync will keep multiple copies of changed files, as many as I select which is now three.
Chronosync will email me if there are any errors in the backup such as the disk being disconnected or certain files that don't work.
Chronosync will verify the backup.
Chronosync copies the files over file for file, it doesn't use some proprietary image file like a lot of high-end backup programs. I want to be able to recover data without having a copy of the backup program itself.
So Chronosync clearly met all of my needs for a backup utility. It worked every night and sent me messages with any errors. It didn't burn up much time making backups and it clearly separated archived files from the original copy. I love it.
However, I realized that I was deleting files on my local disk knowing that they were in the archive. For example, when Michelle and I finish watching an episode of Heroes, I would delete the episode knowing that a copy was still on my backup disk in the _archive directory if I should want it again. With only 80 gigs on my local disk, I didn't want to keep it too big.
If I am deleting the data on my original disk, then that data is not really being backed up. It now only exists on one spinning disk, my external backup disk.
This got me thinking that I needed some sort of raid 1 setup. If I had an external raid 1 drive, I could copy data to it nightly, delete data locally, and know that the archived data still existed on two independent spinning disks. Raid 1 would mirror the data on both disks the minute my backup program sent the data over.
There are two problems with Raid 1, however. First, it's expensive. An external firewire raid 1 enclosure costs at least $100 to $200 without any disks. Adding two reasonable disks costs another couple of hundred bucks.
Second, a raid 1 array still uses only a single disk controller. If that disk controller goes bad, a very rare thing no doubt but still possible, than you could lose the data on both disks. This might be the case if the raid controller card was doing something funky to the disks themselves. I know this can be a problem with a raid 5 array and it might be a problem with raid 1. Mix this with the higher cost and it doesn't seem like a very good deal.
Also, if I am only copying my data over once a night, who cares if its mirrored at the very time data hits the platter? Data will only be copied over to both disks once a day anyway, why not just copy it to two different disks?
So I ran out to Micro Center and bought a $10 firewire cable to hook up both of my external firewire disks, one of them a 500GB MyBook and the other a 160GB Western Digital disk. I set up two identical Chronosync profiles to copy data at 4:15am and 5:15am to each of the two external firewire disks. Now the disks should be identical copies of the same nightly backup, I can safely delete data on my local disk and know it exists on two spinning disks, and should anything go wrong with either of those disks, I can replace it for about $150.
For off-site backups I have two small portable USB 2.0 disks. Every three months or so I will copy over a mirror of my current disk and take it to my safe deposit box at my bank. One portable disk will be at home locally and one of them will be in the box. Every three months I will copy my data and flip the disks.
Really super critical data like my short stories, web postings, favorite pictures, or copies of my father's books will still be stuck in my lifebackup or loaded to my web host provider so it exists remotely on the net.
Overall it seems like a good solid backup plan and it didn't cost me $1000 for a raid 5. All of this can be done with two large external firewire disks and two smaller portable USB 2.0 disks for a total cost of about $500 to $600. That's a small price to pay for a backup routine that should last years and gives me a solid peace of mind.