Mat wrote: > Just an idea: what if both a KVM snapshot and the LVM snapshot could be > done at the same time? It should then be possible to bring the VM up in > a running state.
I guess that might be good in some circumstance, but only when it's ok to get a system in a previous running state. This is not always the case; for example, you would not want a mail server to go back a couple of hours in time (mail in the queue at backup time would get redelivered at restore time, mail deleted from user inbox would reappear, mail delivered to user's inbox since backup would disappear, etc). > My idea is probably trying to address the underlining issue from the > wrong angle. I suppose that if I needed a service to run 24/7 and > downtime was not acceptable, I would run my services in a server farm > with proper fault-tolerance and redundancy, and where database backup > and recoveries wouldn't be done on the VM level. I do not know. Backup is a tough problem, and there is no one-size-fit-all solution. Dumping the file system content and checkpointing system states have been the two most common approaches to date, but they are flawed for many applications. It's all a matter of compromise. But yeah. backup is second best after resiliency, so you may want to invest your time and brain cycles in that instead. Just my 0.02$ ... -- Etienne Goyer Technical Account Manager - Canonical Ltd Ubuntu Certified Instructor ~= Ubuntu: Linux for Human Beings =~ -- ubuntu-server mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-server More info: https://wiki.ubuntu.com/ServerTeam
