NYCPHP Meetup

[csnyder]

On Jan 18, 2008 1:35 PM, Cliff Hirsch <cliff at pinestream.com> wrote:
>
>  Is anyone using Amazon S3 for backup? Instead of getting another
> independent server, setting up rsync, etc. this seems like a good solution
> for backing up a server or sql database. Sure, my ISP does daily backups,
> but having more control by backing up the database off-site would make me
> sleep better.
>

I keep wanting to do this, but several things are holding me back:

1) How redundant/safe is S3, really?

I trust Amazon more than some other large providers, but S3 is a
high-value target and the SLA is for service uptime only. It doesn't
mention the possibility of data loss, corruption, or breach of
privacy... why not?

2) Lack of S3 support in existing tools/kernels

The weird http interface and naming conventions have made it difficult
to use S3 as general-purpose storage. I find it odd that Amazon's
engineers haven't developed a FUSE plugin for the service. There have
been efforts in this direction, but Jungle Disk seems to be the only
one that is decently organized and production-ready.

3) Physical storage gets cheaper while S3 remains the same

...and the monthly cost of storing the data is going to grow as your
backup archive gets bigger. The main reason to use S3 is that you get
(hopefully) redundant, (hopefully) geographically diverse storage
without having to pay for a server, rack space, and sysadmin costs. It
will be interesting to see if the price of S3 keeps pace with Moore's
law.

-- 
Chris Snyder
http://chxo.com/