Best Online Remote Backup Service w/Linux Client?

technocraft asks: "I've been searching for an online service to backup data from my Linux file server and have come up with nothing. For many users, Carbonite looks to be a great solution: Affordable, with 'unlimited' capacity. Unfortunately for me, you can only backup from Windows XP and explicitly NOT from external drives or network mapped drives (like my file server)." Is anyone aware of an online backup services without these restrictions?

Comments

This looks pretty good

[dtfinch]

They use rsync over ssh:
http://www.rsync.net/
Base rate: $1.80/gb/month
Volume discounts:
25-49GB - 10% Off
50-99GB - 20% Off
100-199GB - 30% Off
200-399GB - 40% Off
400-999GB - 50% Off
1TB+ - 60% Off

You get supposedly unlimited storage, and pay for only what you use.
I haven't actually tried them though.

Re:This looks pretty good

[Bogtha]

We're customers of the parent organisation, JohnCompanies, and I can't recommend them highly enough. The thing that makes them stand out in my opinion is the support, you don't get clueless newbies reading from an FAQ like you do with every other hosting provider I've ever encountered.

I've been looking for an excuse to try out rsync.net, it seems like an interesting service, for example it offers WebDAV access, which is built into Windows, OS X and KDE.

Yeah, I know I sound like a fanboy or astroturfer, but I don't care, it's really rare to find a company that actually gets it so right when there are so many incompetents around, especially in the hosting business.

Re:This looks pretty good

[swillden]

That sounds really cool, except I'm surprised at how high the price is. Between photos and home movies, I have about 100GB of data I'd really like to get backed up offsite, and it grows by about 2GB per month. I have some other stuff I want offsite as well, but it's very small.

So, according to their pricing scheme, it would cost me $126 per month to store my data, and the price would increase by about $4 each month as my data grows. Storing my data for a year would cost me nearly $1,800. That probably makes sense for some people, but certainly not for me. It's much cheaper for me to buy a machine full of drives to put at my brother's house and then rsync the data there. Actually, that's exactly what I'm looking into doing -- most of my family has high-speed Internet connections, so we're taking inventory of who has how much data they'd like to back up and how much free space they can donate for others to back up to. After we figure that out, we'll buy additional storage where it makes sense and set up automatic over-the-net backups using rdiff-backup and some scripts.

A commercial solution would be a lot simpler, but at ~$2/GB/month it's priced way too high. For me, anyway. At one-tenth the price, I'd consider it.

Re:This looks pretty good

[cerberusss]

What I like about these guys, is the use of their canary. When they are forced by the authorities to pass your data, they're probably also forced to keep silent about this.

So what do they do? They send out a weekly 'canary' saying nothing happened. If something DOES happen, they don't send the canary.

Re:This looks pretty good

[igb]

Taking a quick look on Froogle, the tape drive alone will cost about $900, and 80gb tapes cost about $70 each.

If you have serious backup requirements, LTO-3 tape drives are about $6000, the tapes are about $80 each, and they hold (in our experience) close to 1TB with compression turned on. You'll need a fast machine to prevent shoe-shining the tape, as you want to drive ~40MBytes/sec into them for best performance. The tapes last for practical purposes indefinitely --- it's a close cousin of DLT, and I think we lost one tape (and that to a mechanical, rather than tape surface, failure) out of a cycle of about a thousand over five years. Many of those had been recycled (ie scratched and re-written) in excess of thirty times. With LTO the most used tape we have is 8 recyclings, but we've had no failures so far.

The casual claim that tapes are old, expensive technology is simply nonsense if you've got 20TBytes of data and you want to keep a weekly copy for three months and a monthly copy for several years. 20-odd tapes times perhaps 30 sets in circulation is 600 tapes, which might cost about $60000. That's $0.10 per gigabyte, at very conservative media costs. A decent six-drive, 72-slot robot might cost you $60000, so that's a further $0.10 per gigabyte amortised over the whole estate.

You can easily split those tapes between multiple firesafes in multiple locations, too.

It's also about 600TBytes: disk is indeed cheap, but if you can build 600TB on multiple arrays for $0.20/GByte you're doing pretty well, and absent some very funky power management that's going run rather hot (assuming that RAID5 is the bare minimum and you can tolerate say 6+1 raidgroups it's still well in excess of 100 spindles). A six drive robot will easily sink 250MBytes/sec, which is at the bleeding edge of very fast disk arrays (that's saturating multiple FC links, unless you're using 4G FC).

What's the benefit of tape? The marginal media costs are low, tapes don't break while they're quiescent, they don't consume power while they're quiescent and they're astoundingly fast.

blacksun

[illuminatedwax]

I use Blacksun which doesn't have the "unlimited" storage you had before, but they are very affordable and offer rsync, ssh, sftp, and the regular linux services as well as the typical dragndrop interface clients. Very nice, and their tech support is helpful and quick to respond!

How do you trust?

[polymath69]

Presumption: Backups are intended for crash or loss recovery.

Presumption: After a crash or loss, you may not have access to any of your own encryption keys.

Query: How can you possibly trust any third party not to take liberties with personal or business information entrusted to their care?

I really think you're better off taking care of backups in-house, along with of course keeping some of those same backups off-site in a secure manner.

DIY Route might be good in this instance

[Kadin2048]

My reaction to rsync.net was similar. I'd really love to try them out, but I can't come up with any really good ways to justify that kind of expense.

If you have multiple locations available for your use -- and if you're a home user, who doesn't? (parents, friends, etc.) -- it's a lot cheaper and you can get a lot more flexibility if you take an old PC, put a bunch of drives in it, and set it up somewhere. Then just have your systems replicate to it at night. (Yes, it's not bidirectional if you just use vanilla rsync, so if you have a lot of file churn you'll need to script something to keep the backup from bloating.)

I guess the 'setup cost' is higher than a managed service like rsync.net, but a minimalist system doesn't even take that long. Install Debian, install rsyncd if it's not already there, and open a port in the firewall for it. (Actually you don't really even need to run it as a daemon, now that I think about it.) WebDAV could take longer, but you'd have to really value your time highly to pay rsync.net's prices in exchange for an afternoon setting up Apache and a couple of cron jobs.

Personally I just have an old 600MHz Celeron machine that I set up with rsync and ssh/sftp with dyndns, and then traded a friend for a similar system that he had set up. I keep his box in a closet, plugged into my router, and he does the reverse. We both get off-site backups, and the only real cost of ownership is the electricity. (And if I was doing it today, you could get one of those routers that can run Linux from CompactFlash and can mount an external HD via USB 2.0 ... I bet you could get the whole thing down to a few watts that way and under $200.)

I am normally very skeptical of the DIY route -- it's tough to compete with mass-production in many instances. But I think that this is one situation where even a low-grade geek can toss themselves together something in a day that will be nearly the equal of rsync.net's service, for workstation-backup use. Hopefully the rsync.net guys will adjust their pricing accordingly and make a liar out of me soon, though.

Jungle Disk / S3

[crt]

Jungle Disk is a cross-platform front-end for Amazon S3 that supports Windows, Mac, and Linux. You only pay the Amazon fees ($0.15/gig/month). On Linux you can mount it directly using DavFS then backup using any software you want (rsync, etc). It supports encryption and caching as well.

Re:Poor man's solution

[cortana]

You might want to take a look at rdiff-backup.

rdiff-backup backs up one directory to another, possibly over a network. The target directory ends up a copy of the source directory, but extra reverse diffs are stored in a special subdirectory of that target directory, so you can still recover files lost some time ago. The idea is to combine the best features of a mirror and an incremental backup. rdiff-backup also preserves subdirectories, hard links, dev files, permissions, uid/gid ownership, modification times, extended attributes, acls, and resource forks. Also, rdiff-backup can operate in a bandwidth efficient manner over a pipe, like rsync. Thus you can use rdiff-backup and ssh to securely back a hard drive up to a remote location, and only the differences will be transmitted. Finally, rdiff-backup is easy to use and settings have sensical defaults.

Basically it's a wrapper around rsync that does two things. First, gives you incremental backups; second, fixes the highly confusing options that control which files are included/excluded in a sync. :)