Even with backups...
There may be data missing, depending on the time of backup. (STBO)
I can imagine that there may be chronological issues too if data is migrated internally to support multiple points of presence.
There’s good news and bad news for customers of French cloud operator OVH. The good is that it has backups of some systems impacted by last week’s fire that destroyed one of its four data centres in the French city of Strasbourg. The bad news is that it doesn’t have backups of some systems impacted by last week’s fire that, …
Backup of cloud systems doesn't seem to get the attention it deserves. No real problem when you're doing it incrementally or simply replicating transactions to a standby system, but if you need to copy and restore TBs of data because of an incident, or just so swap to another cloud vendor you're potentially looking at significant elapsed time and bandwidth cost.
"No real problem when you're ... simply replicating transactions to a standby system"
The real problem is getting it past the bean-counters. "We're paying how much for this? And you want to pay the same again for another one just in case?"
Remember that this was all too often sold on the basis that it makes problems like this go away. Of course all it does is make them just go out of sight.
But the bean counters report to 'The Management' who made the decisions (based on advice from the Techies and the Bean Counters).
And so the recursion continues - until 'The Management' realises that they are in charge and can blame whomsoever they choose, provided the shareholders don't get uppity and blame them. But if they are senior enough, 'The Management' will doubtless get jobs elsewhere with their chums, and their merry-go-round will continue.
>The real problem is getting it past the bean-counters. "We're paying how much for this? And you want to pay the same again for another one just in case?"
"If we only have one and lose it as a result on an unlikely catastrophic event, the company will cease operations and be unable to recover. Please kindly sign here that you are happy to accept this risk".
So backups (where provided) are clearly not off site, perhaps not even out of data centre. Very poor.
One of the problems with many cloud providers is they seemingly purposely produce backups which you can not download, let alone spin up somewhere else if need be.
Fortunately I have managed to work out how to do a backup and restore of a tar backup for centos and ubuntu so I have at least a fighting chance of disaster recovery is my providers backups aren't available. I use them to replicate our live VM's on to a local VM box for testing. In addition I have daily rsync's of a static data backup and replication of data during the day.
erm... it depends if those customers -PAID- for off site backups. AFAIK, those who were paying for the service called "FTP Backup" are safe because those were transferred at Roubaix DC. It's the same drill with every cloud provider, example: Azure Backups, nice. Then you use LRS (locally redundant storage) because it costs less and this happens, you lost backups there as well. "But Geo-redundant storage is expensive!", will say the bean counters...