Amazon says that about 0.07 per cent of the EBS storage volumes in the East Region of its infrastructure cloud are not "fully recoverable" following the extended outage that hit the service last Thursday. The company has yet to fully explain the cause of the outage, but it still plans to publish a "post mortem" on the incident …
Quantify "0.07%". While it looks a small number it could be quite significant if we knew what value the percentage was of.
says (quoting some weird source) that amazon had 102 billion objects on the S3. Use that as a guide for the scale of their little operation.
Goes to show...
Not every cloud has a silver lining
Wasn't the data "In the cloud"? That means it's safe, secure and always available? Doesn't it?
Surely I haven't been lied to by advertisers and marketing men?
Here's what really happened at amazon over the weekend http://www.youtube.com/watch?v=m3wrBFuGK2A
root cause: incompetence
and/or beancounters listening to liars selling snake oil rather than listening to experienced architects pointing out that what can go wrong will go wrong and may therefore need funding for suitable precautions (technical or otherwise) just in case.
If they can't guarentee backup, they have no viable business.
It could be that they do take good backups, but the lost data is since the last backup.
I don't know the RTO or the RPO for the Amazon cloud, so it could be that a certain amount of transaction/data loss is to be expected. Zero transaction loss systems are typically only found in institutions like Banks and then only for payments systems, they are very, very expensive indeed.
The other option is that they've had a failed backup and won't be able to recover the data except from the pervious backup. However this could turn out to be them having had a long time failing backup, in which case they need to be avoided.
Ask anyone who works in backup and recovery or storage in general and they'll tell you that in a large enviornment you always get backup faliures from time-to-time.
Another triumph for The Cloud, where all your data is magically available and backed up. I'm just off to move all my personal data to it immediately.
0.07% - could be a lot more?
0.07% - could be a lot more if it were fragments of files which could now be corrupt.
Staggered by the lack of honesty / information Amazon have provided.
0.07 of a few exabytes is rather a lot of data. That could be a few thousand small businesses that have now lost their primary storage with the associated downtime of recovering from their ONSITE BACKUPS.
Number of affected systems
The Rightscale post estimated that about half a million EBS instances were affected by the outage. If that's the case 0.07% is 350 instances.
Not Ready for Prime Time? The Old Maxim, Back Up, Back Up Still Applies
Given Amazons cloud history, along with it's peremptory transactions with Wikileaks, it is obvious the service is not ready for prime time.
Obviously Amazon can't provide reliable back-up, the need for clients to secure their own data remains effective.
Outsourcing, Virtualization, Planning to Fail
One can choose to build apps on a 100% recoverable platform, with data replication and mirroring that survives multiple simultaneous failures.
But one cannot do so on the cheap.
So long as companies do not value IT, these things will happen.
- Stick a 4K in them: Super high-res TVs are DONE
- Review You didn't get the MeMO? Asus Pad 7 Android tab is ... not bad
- BEST BATTERY EVER: All lithium, all the time, plus a dash of carbon nano-stuff
- Bring back error correction, say Danish 'net boffins
- That GIANT ASTEROID that killed you? Just 'colossal bad luck', old DINOSAUR chap