How good are your backups?

PICT0008We all know we need to do backups. But that is only half the story. Have you actually checked you can read them back if you need to? I have heard stories of people religiously backing up to mag tape every day for years, only to find out the tapes were corrupt and couldn’t be read back when needed.

I checked my backups recently to ensure I could read them back. Here is what I found out:

  • I was backing up my SVN repository on my Mac Mini to a single .tar.gz file which I then copied across onto a USB disk attached to a Windows box. The file had grown unnoticed to >4GB in size. But the USB disk was in FAT format, which has a maximum file size of 4GB. The the file was quietly being truncated to 4GB and I couldn’t uncompress the file when I tried with gunzip.
  • I was backing up from my Windows box to DVDs in ‘live’ format. These were not readable by my Mac mini, which would have been a problem if neither of my Windows boxes were working. I have now changed to ‘mastered’ format, which is readable by Windows and Mac.
  • I was backing up the .mdf file my OnTime bug tracking database. It turns out you can’t just swap one .mdf file for another and re-start OnTime, as you get an internal consistency error. I am now creating and backing up .bak files, which you can restore.

I have lots of redundancy in my backups – backing up in multiple formats (files, SVN repository and Acronis disk images) to multiple media (multiple machines, USB disks, DVD and online). So none of the above would have spelt disaster. But it does bring home the importance of testing your backups from time to time and of having multiple forms of backup. If you think backing up to a single USB disk is enough you should read this. If you are relying purely on an third party online backup service you should consider what would happen if they went bankrupt – not inconceivable in the current economic climate.

Are you relying on a single backup strategy? When was the last time you tested your backups?

19 thoughts on “How good are your backups?

  1. Dan Weese

    I don’t use backup files. Hard drives are cheap. I use MS SyncToy to make uncompressed copies of all of my important files to 2 external 1TB hard drives. The whole setup cost $300 and now I have 3 places for my files and they all stay in sync. The backups run automatically at night so the most I would lose is what I worked on that day if my primary drive crashed.

    1. lostitall

      Local backups are a great strategy and more than what most people do. What happens if you have an incident at your home or office? A fire, electrical storm, flood or theft could wipe-out your computer and your backup.With high speed Internet access online backups are cheap easy and offer some flexibility.
      Free-online-backups

  2. Martin Cote

    It is also very important, if not more important, to have an external backup somewhere. And by external I mean not in the same city as you are. What if your house caught on fire?

    I’m using my MobileMe cloud space for that matter. I upload an encrypted version of my backups to the cloud once a week.

  3. Rico

    At work I’m the guy who works on the backup solution that we sell for $30K+ so I figured I should probably also backup my laptop at home. I rsync my entire home directory to an external USB drive. Unfortunately Apple’s Time Machine app does not work well with encrypted home directories so I had to resort to lower levels. I probably to this once a month. I actually tested the restore and it works like a breeze.

    I also think it is extremely important to have an off-site backup, i.e. to keep a copy of your backups as far away from the source data as possible. What happens if somebody breaks into your apartment and takes your shiny laptop and that nice big, fat external hard dive of yours?

    I thought about getting another external hard drive or large USB drive and keep it at work, but hey, why not store it online?

    So I created another encrypted disk image with my home directory (excluding images/video to reduce size) and uploaded it to http://humyo.com/. They give you a free account with 10 GB space and web access. Easy enough and all I need.

    The three key points about backing up your data are:

    1) Do a regular backup in the first place (automate as much as possible so you actually do it)

    2) Test the restore BEFORE you need it (as you pointed out, can save a lot of trouble)

    3) Keep one backup copy FAR, FAR AWAY

  4. Thomas

    I do three levels of backup:
    1. A laptop which receives the new files by sync. If the PC breaks down, I can still work.
    2. A backup of the most important data on an USB stick. Not likely that a burglar will find it.
    3. A comprehensive backup on an external HDD stored far away. In case of fire.

    Since one of my programs is a file sync and copy software, my choice of tools is pretty obvious. The nice thing here is that I can exclude unwanted files / folders to reduce the backup size and that it’s 1-click.

    And yet: I’m not doing the backup as regularly as I should, I have to admit…

  5. Wille

    I have somewhat of a paranoid backup strategy:
    – Always have my laptop with me, and back that up with Apple’s time machine constantly (onto an external drive I have with me).
    – Periodically copy things to an external hard drive I leave at home.
    – Periodically copy things over to Amazon S3 storage
    – Keep all code in Subversion hosted on Amazon, which in turn is backed up nightly onto S3 snapshots.

    ..I think I’d basically need to have my house burn down, have Amazons data centers burn down and have my laptop bag stolen to loose my data..

  6. Andy Brice Post author

    @Dan

    Do you have an offsite back-up as well?

    Syncing to 2 USB drives protects you again the failure of 1 or 2 drives, but it doesn’t protect you against:
    -someone stealing your computer and backup drives
    -your office burning down
    -a corruption or deletion that gets synch’ed to your backup drives (e.g. you delete a folder and then decide a week later you need it back)

    1. Dan Weese

      Andy, I use Dropbox to sync important files between home and work. It’s not 100% of what I’m backing up simply because that would be expensive. Also, it’s not like you can trust the cloud either. You have to have a multi-faceted backup strategy to make it fail safe. I’ve also been considering buying a USB backup safe. It’s fireproof for 2 hours and with all of the monitored heat and smoke detectors I have, and having a fire department 1 mile from the house, I’d like to think I’d be pretty well covered. There is no 100% though.

  7. Alexander

    I scheduled daily offsite backups via ftps to remote server with raid 1 build on 2×1TB hard drives. All websites are also backed up on it.

    Its capacity will be sufficient to «go back» up to half-year for any project or important document. After that deadline I’m going to move some backups to dvd’s and clean up unnecessary backup copies.

  8. Simon Strandgaard

    I nightly take snapshots of my GIT repositories and make gpg’ed files out of it. This I upload to a private folder on my website. I keep the 7 latest snapshots around.

    Further more I nightly harvest sql files for all my websites and that goes into the gpg’ed files as well. I use ruby mechanize to login and export via phpmyadmin, since some of the websites are on cheep webhosts where there are no other interface for this.

    3 years ago there was a fire in the basement of my appartment, that fortunately was stopped by the firefighters. I had a breakin 5 years ago. Statistically it time again for something bad to happen.

    I’m somehow not satified with my backup solution. It does the job for my source code, but it’s not elegant.

    Also any ideas how to do versioning of multimedia files and proper backup?

  9. Chuck Brooks

    Brings to mind Andy Grove’s comment about only the paranoid surviving. A hard drive failure a couple years ago brought the lesson home. We’ve always kept weekly backups, in various media over the years, swapped through a safety deposit box.

  10. Scott Kane

    I have four sets. Local to a couple of USB HDs I swap in and out, Amazon S3 for the server, dev machine and stuido machine. Off site via another two USB HD’s for critical stuff I can’t afford to lose (source, doc’s et al). This was driven home to me this year when the street my sister lived in burned – every home (along with several thousand others in my region). I think I’ve become paranoid about backups.

  11. Nadya

    Those who use Windows may want to try CloudBerry Backup. It is powered by Amazon S3 reliable and cost efficient storage. What safer place to keep your files than Amazon’s servers? You can download the product at http://cloudberrydrive.com

    Nadya,
    CloudBerry Lab team

  12. SF

    Great that you brought it to peoples attention. Unfortunately, so many lay people just ignore this … and end up losing family photos, etc.

    My development, personal, family and servers get backed up (using SyncBackSE versioning) to a backup drive (on the server). It is all in native data formats.

    The backup drive gets mirrored every night to a NAS running in my workshop/shed about 300ft away on the property. Syncback has email on backup error so I know if it detected something wrong.

    I test recover from server backups and NAS once or twice a month.

  13. Boofus McGoofus

    From http://www.taobackup.com (which is all worth a read):

    The novice asked the backup master: “Master, now that my backups have good coverage, are taken frequently, are archived, and are distributed to the four corners of the earth, I have supreme confidence in them. Have I achieved enlightenment? Surely now I comprehend the Tao Of Backup?”

    The master paused for one minute, then suddenly produced an axe and smashed the novice’s disk drive to pieces. Calmly he said: “To believe in one’s backups is one thing. To have to use them is another.”

    The novice looked very worried.

    You don’t have a backup until you’ve tested it.

  14. Vasudev Ram

    Good subject for a post, Andy.

    It’s amazing, the number of software developers who don’t take backups at all, let alone use the many good backup techniques mentioned in your post and the comments.

    A related point, also surprisingly common, is how often developers lose work (i.e. code that’s already done and is working, or mostly working), in their haste to make changes, or to fix some bugs – I’m talking about when they end up deleting fragments of working code or introducing new bugs into it through their edits; they don’t bother to take local backups, or better, use a source control tool like CVS, SVN, git, etc., to make a snapshot of the files to be modified, before modifying them, so that changes can be rolled back if needed.

    Andy: What’s your solution for this point you mentioned:

    >-a corruption or deletion that gets synch’ed to your backup drives (e.g. you delete a folder and then decide a week later you need it back)

    Boofus: Those Tao / Zen Master stories are good :) See some of the ones about UNIX, here, in Eric Raymond’s book The Art of UNIX Programming:

    http://www.faqs.org/docs/artu/unix_koans.html

    – Vasudev

  15. Andy Brice Post author

    >What’s your solution for this point you mentioned

    Copy a snapshot onto a DVD once a week.

    Having the file/folder in source control also works.

    It also amazes me that there are developers not using source control. But that is a topic for another post.

  16. Dan Hite

    I used to burn DVDs until my storage requirements exceeded what a DVD could hold. These days I have two 20GB external drives that I bought for around $25 each on eBay. They fit easily into a small safety deposit box at the bank. Once a week I swap them (the bank’s on the way to the local movie theater and I like to celebrate my being my own boss but giving myself a couple hours off each Friday afternoon to take in a movie).

    I don’t compress or encrypt my data when doing a backup – I want it to be as easy to access in my backup storage device as my workstation’s HD. I use FolderMatch for copying the files (www.FileMatch.com) which does a fine job.

    My question to those storing their backups “in the cloud” is “How fast can you retrieve your data?” One person told me that while Carbonite seems like a great solution, if you have a large amount of data you need to retrieve it can take a very long time to get it as the downstream speed is fairly limited.

Comments are closed.