Editor's Note: The Motley Crew Collection of Backup Schemes and Devices
Apr 09, 2010, 23:04 (11 Talkback[s])
(Other stories by Carla Schroder)
Full Text Search: The Key to Better Natural Language Queries for NoSQL in Node.js
by Carla Schroder
Back in the olden days tape backups ruled. I never did
like tape. It took too long, it was noisy, and restores were
painfully slow. Most backup software was operating-system
dependent, which made bare-metal restores pure pain. First
reinstall the OS, then the backup software, and then if the gods
were kind you could actually find and restore your data.
This modern era is much better, because we have all kinds of
fast and inexpensive ways to make large backups, and even better,
to make fast restores. I am all about simple and easy. So my
favorite is using plain old rsync and hard disks.
This is my standard rsync incantation:
rsync -av --delete /home/carla/ /media/disk/backup/
I'm fairly diligent about keeping my files pruned of cruft, and
browser caches and trash empty, so I just backup my whole home
directory. -a means archive, or preserve
everything as-is, such as permissions and ownership, and recurse
into subdirectories. -v is verbose, and
--delete means if I delete something from my
computer, then it should also be deleted from the backup.
There are all kinds of ways to refine and pretty this up with
scripts, and to exclude files and directories.
rsync is efficient because after it copies a file
the first time, it only copies changes on subsequent backups. So
your backup drive needs to be larger than the drive you're copying,
but not hugely larger.
Some folks prefer to compress their backups with good old
tar. tar is reliable and easy to
use, and you save a lot of space. But I prefer
rsync because I can read my backup drive just as
easily as the original, so when I want to restore files it's simple
copying, and no fussing with unpacking archives.
The time is coming when I will have to break down and build a
proper fileserver for my home network. But for now I use single
SATA hard drives in USB enclosures, and USB sticks. (Like
Rosewill's external USB enclosures, which cost under $20 and work
just fine.) They're fast and easy, and I can plug them into any
computer to read them. Well, any Linux computer, since Mac and
Windows are so deficient in their filesystem support.
Long-term archiving is still a problem. SATA drives should
outlast USB Flash media. But who knows how long they will last. USB
sticks are nice for short-term project backups, and sneakernetting
multi-gigabytes of audio and image files.
Really Important Stuff
Very important data go offsite to Spideroak.com. They have been
reliable, have a lot of nice remote access and sharing features,
and what I like best--"zero knowledge data encryption". They cannot
see your data because you get a unique encryption key. If you
forget your password, wave bye to your stuff. This is totally fine
with me because I like to think of myself as a grownup, rather than
the mythical dimwitted user targeted by operating system
developers, who think we need to run multiple commands to really
delete something, because deletion is scary and dangerous and maybe
we don't really mean it.
I used to share offsite network backups with friends, some
simple rsync transfers and encrypted directories.
I'll do it again when I can find someone suitable to peer with.
Next week let's talk about what goes into a good reliable
storage server. Hint: RAID 10, not RAID 5.