Jungledisk has been down since Saturday - Mosso Cloud Files Fail
forum.jungledisk.com
forum.jungledisk.com
Sounds like it was a Jungledisk problem, not a Mosso problem.
Personally, I back Anybots up to:
- a 3 TB ReadyNAS in the building (everything nightly)
- rsync.net (important stuff nightly)
- dropbox (important stuff instantly)
- USB drives I keep at home (mostly everything monthly)Personally, I'm working on setting up a local backup and backing that up to the cloud.
Unfortunately my up tube is too small, and the data won't fit. The initial 100GB upload was gonna take 40 days, so I stopped it after 10 with slow internets.
If you have the tubes, I do recommend it though. Its a really solid backup setup.
It was certainly disconcerting to see the failure of a component that was supposed to be the core of my infrastructure. Cloudfiles just moved out of beta about a week ago, and for growing pains the outage was pretty long.
According to their status updates, data integrity was never at stake, just availability. CDN files were always accessible.
I'm relieved to see that the Jungle Disk integration seems to be to blame for this, since this hints at a one-time peak load event rather than an overall infrastructure failure.
I think the ease of use of the CDN and low latency of delivery for many small files is a big win. Is anyone else here (thinking of) using them?
So while I may be biased - I'd like to point out that a bug in a end user product brought down their entire cloudfiles platform and that the same bug also effected S3 and had no noticeable effect.
I use S3, but not JD, so I don't keep up with it much. I thought that when they added Mosso, it was in case S3 went down -- am I misunderstanding how it works?
Yes. With Jungle Disk you pick where you want your files stored.
As far as I can tell Jungle Disk added support for Mosso simply because Rackspace bought JD and were embarrassed to not be using their own cloud storage system.
Even with multiple data centers, things like this can happen if bad data gets replicated to multiple data centers.