Is anyone writing backups to cloud storage without the Cloud Connect product? I need a reliable way to get my data offsite into inexpensive long-term cloud storage like S3 or Glacier, but I want to do it directly without relying on a third-party partner.
I've tested backing up to virtual tape using Amazon's VTL solution, but because Veeam requires writing the entire backup set to disk before performing a restore, recovering even a single file or VM would require downloading 4-6TB of data, making it prohibitively expensive and creating a significant delay.
One potential solution I can think of is to host a repository on Amazon Storage Gateway Cached Volume (http://docs.aws.amazon.com/storagegatew ... cepts.html
), but I have a number of concerns and I was hoping someone here had either looked into this as well or even implemented it in some form and could comment on their experience:
1) How would you deal with maintaining restore points over the long-term (5-7 years)? Would you continually write them to a single massive volume, or create smaller volumes and swap them out as they fill up? Does it make sense to take Amazon EBS snapshots at intervals and then delete the older Veeam recovery points after they had been snapshot?
2) Are snapshots even practical? The cost to store snapshot data is >3 times the S3 storage cost ($0.095/GB vs $0.030/GB)
3) Is there any way to move the data from S3 to Glacier to reduce cost ($0.010/GB vs $0.030/GB)? I don't see anything in the documentation indicating that it's possible to move Gateway data to Glacier.
4) In the situations above, if I roll off older restore points by swapping out volumes, or snapshotting, or moving them to Glacier, how does Veeam deal with the disappearance of these recovery points? How can I find them in my inventory years from now when I need to restore a file or VM from a specific date or based on other criteria? Do I have to keep everything online all the time just so Veeam doesn't lose track of it? Is there any way to tell Veeam that the data has been archived, and then locate that archived data years later?
5) For performing restores would it be beneficial to build a Veeam proxy server in Amazon EC2 that would have faster access to the data in S3? If I did this, could would it be possible to restore a VMWare guest directly to EC2? Is this a practical way to restore files or VMs, either to an EC2 instance or back to my local LAN?
Am I overthinking this? Is is better to just use Cloud Connect or the Amazon Virtual Tape Library? If you've used either of these, Id like to hear about your experience