Amazon EC2 + EBS:: Regular backup plan?
I have read some articles about amazon ec2 instances and ebs. I found out that, the data on the instance are not persistent, which means, if the hardware fails, instance shut down, I will lost my data. Hence, what is your backup plan for this case?
Q1 How do you backup the necessary files and database (mysql)?
Q2 I created/mounted an EBS volume. What should I do with it? (How big should I assign the Volume, by default, I assigned 1 GB, is it enough? Or is it extensible?)
Q3. Whats the snapshots in EBS? (for backup?)
My OS: Amazon Linux Using ec2 instance, installed httpd and mysql
EBS volumes are persisted when your instance fails:
"Amazon EBS volumes are off-instance storage that persists independently from the life of an instance." see EBS Information
"Local instance storage" is probably what you're thinking of - and that is lost when the ec2 instance is terminated. I use local instance storage to handle data that can be rebuilt from elsewhere. It can be very handy as a fast cache of data - but you must be able to rebuild it.
Q1: Either use Amazon's RDS service to backup your database regularly and automatically, or run mysqldump on a cron job, compress + encrypt your backup file, and perhaps push it to S3.
Q2: You can snapshot an EBS volume, then create a new volume from the snapshot and specify the storage space (effectively expanding it).
Q3: Snapshots allow you to roll back to a point in time - so before you do potentially breaking upgrades, you could snapshot to ensure you've got a rollback plan. You can snapshot as frequently as your data/service demands.
On top of what Steve said, EC2 instances have the same backup needs that any server has.
There is not a single pre-canned backup solution for all EC2 instances. A web server, database server, file server, domain controller, dns server, etc all have different backup needs.
It doesn't matter if your storage is persistent when someone accidentally runs a DELETE FROM without a WHERE clause, or your website is compromised or anything else happens that doesn't cause "physical" data loss, but makes the current data no longer useful.