Last September I wrote an article about backing up a Mac. I’ve received several inquiries about how best to back up while using Time Machine.
I find I have two distinct needs for backup. The first is total disk loss. If my system disk fails, I need to get back up and running with a new disk. If that disk is pre-populated with the OS and all my data, then I have an extremely fast and reliable way to recover. I consider this part of my backups absolutely essential and I use SuperDuper for making and updating these disk disks.
Creating a bootable snapshot requires a spare disk. My dual G5 had extra drive bays so it was easy, buy and insert a spare disk. When I switched to the Intel iMac, I connected those same backup disks to the iMac and laptops via my Wiebetech ComboDock. That worked but wasn’t very convenient so I picked up an external dual drive Firewire to SATA enclosure with a spare hot-swap drive tray.
My other backup need is file recovery. If a file is deleted, incremental backups allow recovery from a specific point in time. A bootable snapshot can only help if that particular file existed when the snapshot was taken. Fortunately, the few times when I really needed to recover a file, that was the case.
Time Machine fills a void in my backup regime and works out perfectly with the dual disk enclosure. One disk stays in the enclosure and is my designated Time Machine disk. Time Machine manages incremental backups automatically. I maintain my bootable snapshots using two disks and the other drive bay. Every other month I swap out the snapshot disk and store the eldest offsite.
What’s your opinion of Drobo (http://www.drobo.com/). I think it looks like a good technology, but is currently $$$. Version 2 is rumored to have NAS capabilities.
It got somewhat excited about it when I first saw it, quite a few months ago (last year?) when it was announced. I can’t recall exactly why I wasn’t completely “sold” but my reason today is simple: ZFS.
We just built a FreeBSD 7.0 server on a HP 320S with ZFS and 14 SATA 750GB disks. We configured it with raidz and Monday we take it to the data center. We have a bit more testing and might wait until 7.0 goes RC before we press it into production.
In any case, you can guess what OS my file server will be running in a couple months. It’s currently running FreeBSD 6.3 with geom_mirror for RAID. ZFS raises the bar and finally delivers on the part that’s been missing from Redundant Array of Inexpensive Devices for many years.