I need to backup a few files, no more than 1Gb initially. These files represent the mission critical data for my startup. If I lose the AWS machines on which the live data is hosted I need to be able to restore the data to new machines with no delay, which puts me off Amazon Glacier as the blurb I have read so far is vague in terms of restoration SLA and certainly mentions hours, whereas I need minutes. Should I be using S3? What are my options? I intend to keep daily backups going back 5 days, weekly for 2, monthly for 2 so after two months circa 9Gb total.

  • if you need minutes for restore, why are you backing up online at all ? Offsite physical backups rotated out would seem more logical ? – Sirex Jul 22 '14 at 0:03
  • Physical? My data is on AWS VMs, seems more logical to find a cloud based solution, – Myles McDonnell Jul 22 '14 at 21:14
  • sorry, totally missed that in your post ! – Sirex Jul 22 '14 at 21:47

S3 is an excellent option. I currently use it with several scripts to do backups for systems files, docroots and database dumps. Some pointers would be:

  • use bucket expiration policies when syncing backups to your S3 bucket's different prefixes (daily/, weekly/ and monthly/) - sync a date named directory and let S3 take care of the retention policy
  • also have a prefix called last-backup/ with no expiration policy and do not sync to a date named directory. If something is wrong for a long time and all the other backups expire, this one will not
  • if you want the beginnings of a DR strategy, also sync to another bucket in another region. I use us-east-1 and us-west-2 because they have the same cost

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.