Risk management here is the key. In the days leading up to Hurricane Katrina and when worrying about Hurricane Wilma, folks in the states realized that having one data center in Miami and another in Houston wasn’t enough coverage for them, because one storm may have knocked them both out.

At Pythian we have daily scripts that check to make sure the backup was succesful. We also discuss the backup *strategy* including whether there is an offsite copy, every 2 months, for each client. This way everyone is totally informed and there is never a problem of “I didn’t realize we weren’t backing that up!” (it’s also handy to see the snapshots in time of how the backups are growing, and how long the backups take).

Amazon S3 is a convenient way to store backups; we’ve developed a system to be able to store and retrieve backups based on the backup date (S3 doesn’t use traditional folders, so the retrieval can actually be rather tricky).

Good advice.

I think it was AST who said a truck full of tapes running down manhattan has more bandwidth than your fastest network. That was written quite a few years back, but may still be correct. Sure, a few GBs are fine; some dozen also OK. A few hundreds – may be a problem; may take quite long to download/upload again. Taking the tape with you by car may be faster

Regards