backup strategies?

Hi everyone- for those of you who keep regular backups, I’d love to hear what/how you do that.
My challenge is that I have 2 machines (work MacBookPro, personal Powerbook) and separate drives for media (I keep my iTunes library off the machine as it is too large).
I have both USB, FW400 and FW800 drives and interfaces, but I’m thinking about going to SATA and some kind of RAID configuration for the next step. If anyone has any recommendations (Sonnet? FirmTek?), I’d be delighted to hear them.

7 comments on “backup strategies?
  1. There are no good turnkey solutions. It’s a huge unmet market, albeit where half the marketing might be convincing users of the need.
    I think the way to go is to grab the easiest-to-configure version of BSD or Linux, build a box-of-disks, and figure out some decent way of pushing the bits from the computers you use to the BOD.
    Of course, the job isn’t really done until the bits are remotely located…

  2. Fazal Majid says:

    I used to have a 1TB (2x500GB RAIDO0) LaCie FW800 drive, but it was quite noisy and I would keep it off most of the time, and thus not get in the habit of making regular backups.
    I bit the bullet a few months ago. I bought a Sun Ultra 40 M2 workstation at a discount using the Sun Startup Essentials program. This box has a dual-core AMD Opteron and will take up to 8 internal SATA drives. It is also very quiet (since I keep it in the room next to my bedroom, I can’t have a noisy one otherwise I wouldn’t be able to sleep) and it consumes only 160W or so fully populated with 750GB drives.
    I set it up with Solaris 10 Update 4 and 6 drives in a ZFS RAIDZ2 (dual parity) configuration. I can survive 2 drive failures before my data is at risk, and ZFS has snapshot capabilities that make it possible to go back in time, while only consuming the minimum amount of disk space to keep each version (by not duplicating unmodified disk blocks between snapshots). I can’t reiterate enough how amazing ZFS is in terms of power (unlimited snapshots, built-in compression, virtual disk devices), flexibility, data integrity (it uses checksums through and through so even if a disk or controller is experiencing silent data corruption, your data is not impacted). The most remarkable thing is how ridiculously easy it is to administer with just two commands.
    That box is also my personal mail server and RSS reader, so my email is automatically backed up when I take my nightly ZFS snapshot. I could take more frequent snapshots, but that would be overkill.
    I wrote scripts that essentially rsync key files from my desktop PowerMac G5 and my MacBook Pro to the Sun. Tim Bray has an essentially similar scheme documented here:
    http://www.tbray.org/ongoing/When/200x/2006/01/31/Data-Protection
    except that rsync is much more efficient than the tar-based scheme Bray recommends. Over gigabit Ethernet, a 350+GB rsync backup takes less than 3 minutes to complete. the hardest part is making a thorough inventory of what you need to backup. Because it is all scripted and painless, I am now much more thorough about backing up.
    Now, since the server is running 24/7, it can also rsync across the Internet to an old PC I keep at work with two 400GB drives in a non-redundant ZFS RAID0 configuration. This is less redundant than my home setup, but sufficient for disaster recovery purposes. The US is a broadband hinterland, but even with our poky DSL, you can still backup about a GB or so per day if you let it run between midnight and 8AM.
    Obviously not everyone can set up and administer a home server, but there are many packaged solutions that have most of the same capabilities, like the Infrant ReadyNAS, or the FreeNAS (FreeBSD based) or OpenFiler (Linux based) storage appliance distributions. Like Solaris 10U4, the better ones even support iSCSI, which allows your networked storage to be mounted as if it were a SCSI drive (OS X 10.5 Leopard will have iSCSI support).

  3. I back up documents and pictures on my Dreamhost account using subversion (which is a piece of cake to set up on DH) – absolutely love it, and it allows you to sync your stuff between and update it from different computers, not to mention the rollback functionality.
    Music and movies are backed up to a local 500GB HD once in a while (too heavy to do this over the network all the time).

  4. yongfook says:

    I just wrote an article on my blog about the solution I’m using. For small, critical amounts of data it is an excellent method. Not so great for massive amounts (e.g. backing up all your music etc).

  5. Kevin says:

    I have forgon backups – gambling with my data is much more fun, and as I discovered a while back when one of my computers finally pooped out, its more of a relief than a stress. I found that almost everything I had on that computer was not really life threatening to lose. Sure, a few things took some time to rebuild, but even then I noticed new things in the process.
    I say live wild, screw backups – nothing that really matters is on the computer. As long as you have your brain in-tact you are OK.

  6. Peter Adams says:

    Gen –
    The most important thing about backups is to ensure that they happen consistently and frequently. Any solution that requires you to plug something in, or turn something on is inevitably going to result in you forgetting to do it.
    The solution that i use to get around this problem is to have a file server attached to the network that I can backup to. I use an xserve as the server but you could just as easily use a Linux box or a Mac Mini (which is what i would recommend to start if you don’t need a lot of storage). A have 2TB of disk attached to my server in a RAID 1 configuration so hat all data is completely mirrored on to two disks in case of drive failure. This is critical as hard drives do fail.
    I went with a Mac based file server because I was backing up Macs exclusively and didn’t want to spend time fussing with Linux.
    Once you have a file server in place then you have choices about what software to use to move the data from your mac to the server.
    I’ve tried a few of these and about once a year i get tempted to try the new new thing. I also agree with whoever made the comment that there really isn’t anything perfect out there.
    I tried all the Mac shareware (deja vu, etc.) and found that it was really inefficient in terms of managing hundreds of GBs of files. They also suffered from either one of two problems: 1) most programs wanted to copy all files to the server regardless of whether they had changed or not and 2) their was no way to restore a version of a file from a particular date.
    I toyed with subversion as another commenter suggested but it grinds to a halt when dealing with lots of large files.
    Then I discovered rdiff (http://rdiff-backup.nongnu.org) and have been very happy with it. It’s one part version control and one part backup software.
    rdiff is an open source project written in python. It installs with MacPorts and provides you with two things: 1) true incremental backups (only copy files that change) via rsync, and 2) file/collection versioning (i.e. restore the version of this file that existed on this day).
    The combination of these features is killer.
    I install rdiff on each mac and setup an automated cron job to run at 3am each night. My backup takes about 30 minutes to run.
    Now all i have to do is leave the mac on while i go to sleep and rdiff does its thing at night backing up changed files over the network to the file server.
    Hope that helps.