Ask HN: Help me fix your backups

Right now about a quarter of respondents have 1TB or more, so while I'm not in the majority, it seems I'm in good company. 47% of respondents seem to be in the 100GB-1TB bracket as of right now.

Here's the breakdown of my data (largish categories only)

==BUSINESS CRITICAL==

* 5GB of codebases I've committed to in the last 3 years.

* 50GB worth of software. Mostly development tools. Five different versions of XCode, I actually use most of them every day (scary, I know…)

* 50GB of VM images I use for development. I could rebuild them, but I really don't want to.

* 80GB on my boot volume. Stuff I would backup if and only if I want a live, bootable backup. Save me a lot of time during a restore.

* 5GB in ebooks; primarily used for reference / searching

* 20GB in music

* 20GB in scanned documents, mostly financial and legal paperwork, contracts, etc.

* 15GB of e-mail

==OTHER THINGS AVAILABLE NOWHERE ELSE==

* 250GB of samples/patches/instruments that I've built over the years and songs I've written (I play keyboards for fun). Not business critical, but I'd be sad if I lost it.

* 20GB of backups from computers I used to own growing up. The nostalgia of that game I was coding in middle-school, etc.

* 100GB in learning resources I've collected over the years that are out of print or otherwise unavailable anywhere else. Not useful to me (anymore), but sharing them with others is valuable to both me and them.

* 20GB of home video

* 13GB in personal photos

==THINGS I DON'T WANT TO DOWNLOAD AGAIN==

* 375GB of Apple Development videos. I refer to these fairly regularly. I typically watch a few each day to keep up-to-date.

* 20GB of Steam games

Looking at this list, sure, I could live without some of it. But a lot of it is pretty important, and it would be nice to back up most of it.

Backup the whole thing on harddisks and store one at your bank (if they do have that capability where you live) and another one at some relatives place. Encrypt and seal of course.

Update that with "slow" data every month or couple of months. Decide what needs to be backupped daily/weekly and pay some money for that to do it online. How much would it be worth to you if you lost it in a second?

Thanks for the recc. Just spent some time reading the Tarsnap website. It sounds to be a very competitive solution, both in terms of operational transparency (detailed, no-gimmick website) and the layer of security.

Hope it's not a stupid question, but how do you deal with the "bus factor" (a.k.a. "truck factor")? In other words, if the service relies on encrypted S3 backups and the founder exits for whatever reason, is there a contingency plan?

(http://en.wikipedia.org/wiki/Bus_factor)

One thing to point out: backups in the cloud are useless if you don't have internet access when you want access or to restore data.

If you want to fix my backups, make the data readily accessible when there isn't a net connection between me and the service.

One time this usually bites is if you just (re-)installed a new OS and you need the network driver, which is only available online. I always keep a backup of the drivers on my external hard disk for this situation.

Another service I worry about is Gmail and other Google apps. With email you can easily download the several GB of email onto your hard disk, thus keeping a backup for when you or the service are offline.

Dropbox solves this nicely: it's a local folder that's synched to the cloud. If Dropbox is down or I'm offline, I still have access to my data.

I'm not the OP but I'll add my datapoint here. I have that just because of digital photography. My current 12MP camera fills up a 16GB card easily if I am covering a 2-3 hour event for some friend. That adds up fast.

Right now I do a versioned rsync to a server I host myself that has a (just upgraded from 1TB) 4.5TB RAID5 array. That was ~400$ in disks when I upgraded. Even at the current ~1TB usage that would be 3600$ per year in tarsnap storage alone, without the bandwidth cost.

What I'm still missing is a way to backup that server. I could just collocate a second server for a fraction of the tarsnap cost, and I don't really want to be backing up into physical media (tapes or drives) and shuttling that to a deposit box. What I need is tarsnap only 10x cheaper. I'll accept a lower redundancy than S3 for it and especially a lower turnaround time for restores. S3 seems to be built for on-line transactional stuff not backups so is a poor fit for these services. Even the reduced redundancy version only seems a marginal difference.