Ask HN: How do you Backup your Personal Files?

25 points by davidbrent ↗ HN
Call me a child of the American media's fear tactics, but this well publicized story of Matt Honan's hack has me questioning my backup techniques of personal files (pictures, music, important family legal docs, tax data, etc...).

Currently I have one hard drive on my home Windows 7 desktop machine, with dropbox sync setup to mirror certain folders and then I use Google Drive for most of my pictures and media sorts. Now I am thinking I need a local option, something fully under my control (and unfortunately my responsibility).

With so many options out there, I wanted to see what the users of Hacker News were doing with their personal files, outside of the plethora of cloud solutions. Thanks!

55 comments

[ 4.6 ms ] story [ 70.1 ms ] thread
Apple Time Machine and two external 2 TB hard drives for the iMac. Additionally, a daily backup backup of my Macbook Air to a 128 GB USB flash drive, which I copy over to the external hard drives once a week.
Mine is similar: Time Machine all the time, and a periodic drive image using Super Duper to one of two external (actually enclosure-less internal) drives, one of which is always offsite.
What will you do if your house burns down?
My first line of defense: Dropbox

I also have CrashPlan running in the background but hardly ever have to use it

I have an Apple Time Capsule as well but, like CrashPlan, I try not to have to use it and it's more of a worst case scenario backup.

Dropbox all the way.

rsync cronjob to my external 8TB :)
What will you do if your house burns down?
I don't know why the comment below this "What will you do if your house burns down?" is marked as dead ... it's a very valid question.

Backing up onsite is almost as bad as no backup at all. You are protecting against hardware failure ONLY.

If your house burns down you are screwed. Also, if someone breaks in and steals your computer, do you think they aren't taking your external drives as well?

Offsite backup is mandatory if you really care about not losing your data.

rsync to RAID 5 externals.

Crashplan running in background.

Google Drive for Docs and working files.

Github & other git services for projects.

I have a cheap box at a webhosting company with ssh access.

I use svn over ssh and check in all of my person stuff there.

This is really nice because I can share linux dotfiles, etc. between my work and home machines.

* Crashplan, family plan that has unlimited backup for 5$/mnth

* Drobo FS for local backup, can go upto 5 HDDs as RAID.

Over 400GB backup on external and crashplan, these include all my personal files and photos throughout my life.
- Dropbox for documents and archived log files.

- Photos and videos are backed up to a truecrypt drive manually using FreeFileSync twice a week or so.

- Backblaze running constantly.

- Pogoplug (+1 external drive attached) offsite at my parents house with archlinux installed. Eventually I'll attach another drive and have rsync mirroring them using a cron job.

Time Machine via a drive hooked to an Airport Extreme along with Backblaze.
I'm finding I generate very few "personal files" these days, compared with, say, ten years ago. Most personal files are produced/archived with Google Docs.

Yes, that is scary.

I also have 100 GB or so of archived photographs and videos that I back up with a mixture of Ubuntu One (videos and assorted files from 10 years ago) and Rackspace Cloudfiles (photos). I wrote my own Python scripts for backing up and verifying photos on Cloudfiles.

I've also been working on less files locally. Everything is sitting in Google docs. Or worse, I've mailed it to gmail and it sits there unread.

I rsync between my two nix boxes. My wife's Win box I do every now and again, mostly video and pictures.

Drop box for scripts I need across the nix boxes.

And I do a manual copy to an external, my folder structure is the same on all boxes but I like to watch this one.

I know you are looking for a non-cloud option, but Tarsnap (an excellent service by HN's cperciva - http://www.tarsnap.com/) is impervious to the sort of issues raised by Matt Honan's hack, as long as you properly secure your private keys. Of course it is still susceptible to the other weaknesses presented by the cloud, which is why I also use an rsync'd external drive.
Same here. Tarsnap + external drive, with the occasional burning of DVDs for Very Important Files™(with parity data).
Hourly remote-initiated rsnapshot to MyBook Live NAS on internal network.

Then the NAS performs a nightly sync of selected folders from the latest hourly backup with tarsnap, keeping backups from the last 7 days, last 4 weeks and last 3 months.

I'm a Linux user. I have a laptop that's my main machine and a fileserver for storage. I try not to keep anything on the laptop, but have my programming projects & config files on it (so I can work if I'm away from home).

The fileserver is where most of my stuff lives; it gets NFS mounted to the laptop. It's got two drives as RAID1, and backs itself up to an external hard drive using rdiff-backup. It also backs itself up offsite using duplicity to a server on another continent. Everything's encrypted, so if anything gets stolen I only have to worry about the material loss.

I've had certain personal data wiped from a hard drive many years ago, so I am a bit paranoid:

USB Thumb Drive for active files, 2 copies on External Hard Drives for long term storage (which I migrate once a year to new drives), and two physical copies of critical documents (legal, tax) kept in two locations (usually my parents house and a safe deposit box).

I've spent a lot of time thinking about this... I don't trust Drobo/Synology etc because if the unit dies I need another one to retrieve my data. Jungledisk adds up if you have 100+GB, Backblaze is the best of these services, however they are proprietary, and imagine if either were cracked and your life ended up on on a torrent server somewhere... In the end I decided upon a HP Microserver running Ubuntu with disks in RAID5 + a cheap UPS. I then have an rsync cronjob on my laptop. I tried FreeNAS but it was slow. I may consider an offsite backup of the Ubuntu server (critical files (photos etc)) to AWS with Duplicity and PGP... So far I feel the safest and happiest I've been in awhile about my backup situation.
Personally I have been very lazy with backups, but I take important documents to usb stick which I carry with my keys everywhere I go.

Though I should probably start backing up my personal hobby projects, because I have lost ton of them over the years, because I just kept one working copy of them, but I have been lazy to great a backup plan which I would actually follow.

First line of defense is a Time Machine backup to a 1TB drive connected to an Airport Extreme. No-brainer if you're on a Mac, in my opinion. Tip: excluding my Chrome profile from the backups reduced the size of most backups by an order of magnitude.

Second, CrashPlan. It's the only decent solution I could find with support for backing up network drives without being prohibitively expensive for large amounts of data. The client is a bit resource hungry, but I'm hoping the situation will improve as soon as the first run is complete.

I also have a Dropbox account which I mainly use for syncing and sharing, but I've included the most important files there as well for extra redundancy and ease of access on other devices.

In addition, all my code is pushed to at least one remote repository, either on Github or a server. I also run my own mail server which is rsynced to another server, so there's at least three separate copies of my mail folders.

I tried crashplan on my macbook air. The big thing I'm noticing right now is it takes freaking forever to do the initial backup. I'm really hoping that subsequent backups don't take nearly as long.
The initial backup can be long yes... I usually leave my Mac always ON for a couple days when that initial backup is happening. After that first backup is done all the others are incremental and usually fast... Unless of course you just created a multi gigabytes movies that need backup.
I have almost the same setup as you, except for the mail system: I am just using either GMail or iCloud server for that. TimeMachine and CrashPlan work really well and do their job totally in background without any visible impact on my work. I really like CrashPlan and went initially for a 3 years unlimited plan: you can get a really good deal usually around black Friday where they drop their yearly unlimited prices.

All my macs are now setup with CrashPlan.

I just keep everything I care about in my Dropbox. I use enough machines on a day-to-day basis that this leaves me with a decent local redundancy, and having it all on Dropbox means that I'll have access to it even if all of my machines go down at once.
I fear Dropbox may be more like RAID than like backup: what happens when a file is deleted, causing it to be removed from all of your redundant machines?
Also if one machine is not updated (due to no network, for example), the file still exists there.
I've been wrestling with this same issue, ever since I read Google's EULA. I'm thinking of using Carbonite, which no one's mentioned yet.
CrashPlan.

Maybe this post should be converted to a poll, though?

Rsync cron to my RAID 6 NAS box. Photos get sync'd to 3rd party service. Thumbdrive for critical passwords (keepass).
I have some items that I would not like to lose - photographs of my son.

These are copied to multiple machines, in different houses. They are uploaded to 2 different providers (facebook (natch) and dropbox.)

I also create rars with par2 redundancy data and burn them to good quality CDs which are stored in tyvek sleeves in a firesafe.

That's perhaps a bit over the top, but I really don't want to lose those photos.

Passwords and serial numbers are printed out and kept in the firesafe. That's perhaps a bit insecure, and I need to arrange an "in case of death" list.

Everything else used to be rsynced to a different machine in the same house, but now it's time machined.

A firesafe is a great idea, and I might follow your example. Do you know if the temperature inside the firesafe remains low enough so that things like CDs and hard drives don't get corrupt? I know that they work for things like papers, but are they deemed safe enough for electronics?
Arq backup works great. It does encrypted incremental snapshot backups to your personal S3 account (unlike most services which hold your backups hostage). It has an open-source recovery program. It has a documented format.

For local backups, I use a poor-man's Time Machine — rsync with hard links. It's how Time Machine works under the hood anyway, and I used it long before Apple baked it into the OS. Works great.

The OP uses Windows, though, so I honestly have no idea what equivalents to these options exist which would work with, e.g., NTFS.

Time Machine also uses directory hard links for when an entire tree is unchanged, as a pretty big optimization.
Mark me down as another vote for Arq ... total set it and forget it setup. Very easy to use and inexpensive option for offsite backups.
1TB Touro Mobile USB, always sitting on top of my desktop. I just have to do a manual backup every so often using Sync Toy.

I also use Dropbox etc free.

I have a 1TB external drive, and occasionally plug it in and run a script that calls `rdiff-backup --exclude-other-filesystems foo /mnt/backup/bar` for multiple values of foo and bar.

It's not complete enough that I could wipe my root partition, restore the backup and start running again, but I'm not going to lose anything important if my laptop gets stolen.

I don't use cloud backups, mostly out of laziness. (I would have to think about what is actually worth backing up, and organise things so that it's easy to back those things up and not other things.)