Amazon’s S3 and consumer storage

This weekend I looked into Jungledisk as part of my interest in Amazon’s S3 and related products. I’m fascinated by the potential to go serverless: to essentially rent out a box and storage as you need it, without having to pay hosting fees, and home use implications.

So anyway. Right now, S3’s prices are:
$0.15 per gig per month of storage
$0.10 per gig of data uploaded
$0.18 per gig of data downloaded

So I thought about this: what would it cost for me to backup a DVD’s worth of data?

For normal 4.7GB disks, it turns out to be $1.18 to upload and store it for a month and then $.71 for every month I leave it up there.

With the price of blank media, I’m better off burning it to media and storing it. Or use a hard drive.

But here’s where this gets interesting. I’m entirely paranoid about backups, but I’m really bad about actually doing them. I try to do a whole rotation, keeping incremental/full backups on a rotation and what not, but I don’t in practice do a good job. So what I do in addition is run two drives in RAID-1.

The only think I’m really paranoid about, though, is the writing. My whole backup paranoia comes from college, when I got lazy and lost about four months of stuff, and restores failed. And that’s only a fraction of my stuff.

And there’s where this potentially shines: incremental backups.

1) Sign up for the storage
2) Map it to a drive
3) Pick your favorite backup utility and set it to go at the one directory I care about on a schedule

Then each day or so, it goes through, sees what documents have changed, and updates them. I get charged some tiny amount for the bandwith burned that day, and at the end of the month, I have to cough up a couple of pennies, because the actual amount of data that changes day to day is tiny.

No having to stash some DVDs at work in case the house burns down. No rotation schedule.

That’s ridiculously cool and convenient.

Linus Torvalds once said “Real men don’t use backups. They just submit their work to a public FTP.”

That’s almost exactly what we’re talking about.

The really interesting thing, I think, is extrapolating out — what happens as encrypted, secure online data storage is nearly free and you can set up your own virtual server to do things with it?