Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I have a serious problem with the proposal offered in TFA and most variants in this thread: it encourages bad backup policy. Critical data needs three (or more) backups. In my experience, one cloud provider can only ever be considered as being one backup no matter their technical architecture.

For example, you can't (and in the broad "you", aren't even qualified to) audit the provider for data-loss or SLA-impacting SPOFs. You just have to assume that they're there. You also take on non-technical failure modes: the provider can go out of business, get bought by an uninterested owner (think Delicious), change focus (Google Reader), or experience myriad other problems.

FWIW, the best success I've had with helping others switch to good backup policy happened once it became economically feasible to buy a new laptop and two bus-powered external drives as big as the laptop's internal storage. The externals become bootable backups, one of which can be preiodically rotated to an offsite location. This has prevented severe dataloss events for myself and others far more times than I care to count now.



FWIW, I'm more technical than a typical user might be but I'd be very intrigued by a provider that used multiple cloud service and gave me the keys to the underlying S3 buckets (or whatever) so I could independently verify. That's probly a more valuable idea to a company selling backup to nerds though, I guess.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: