geek shiny

Amazon S3: How to back it up? (also S3 and MogileFS)

So, there is much talk about using S3 for backing up your production data, but what if your production data is *in* S3?

S3's resilience and availability is oft touted (although with some hyperbole, three 9s could still leave you down for a business day without recourse) however user, coder and admin error occurs and, perhaps, a get out is needed if S3 does go down for whatever reason.

So, do any of you have precious things in S3? How to you keep them safe and secure? Inquiring minds want to know.

As an aside, MogileFS and S3 seem to have a certain affinity. Has anyone tried to put together an interface so that applications could be coded to use either at a whim? Seems like it shouldn't be too hairy, but then I'm an admin, what do I know about coding. :) If that were reasonable, then would syncing a local MogileFS install with your S3 buckets be a reasonable way of keeping backups, and how would you handle revisions of files in that instance...
Why only use the one cloud? I also backup my files from Amazon S3 to the Rackspace Cloud Files service.
Re: Cloudy
Well that works just fine too! :) By what process are you doing that?
[grin] A quick bit of Wikipedia-ing cleared up some of my confusion. And led to the realisation that I am, in fact, an s3 user by accident - as a result of work I use BaseCamp, and my photos are hosted by SmugMug, who use s3 themselves.
Hehe, perhaps. Your problems come very reasonably priced, however.
For backups, sure, but I wouldn't want to run my live server off them.
I may not be terribly familiar with Amazon's offering but arent you paying a lot to backup that data somewhere off Amazon S3? Are there any partner companies with Amazon whereby you could transfer the data as backup but not have to pay bandwidth?
Well, you are paying, but not really a lot. Amazon's bandwidth charges are really very reasonable unless you're talking about massive data changes.

I'm looking at for backing up my datacenter servers into S3, I wonder if it could be used the other way...
Although you make a good point. If I'm backing up to prevent the human element then another bucket within S3 would be perfectly acceptable as storage and I could avoid the transit costs. Food for thought.