Currently we run rsnapshot on a couple backup nodes for all of our management services clients. A huge selling point is how good our backups are, and some people just pay for us to back their servers up, but we think they could be a lot better.
Reasons we love rsnapshot:
We're able to connect from the backup node to remote servers over SSH. No additional daemons need to be run on the endpoints.
It uses rsync incremental transfers and uses hard links to save on both disk space and network utilization.
Reasons we don't love rsnapshot:
Jobs can not run in parallel. Encryption is going to be the bottleneck, I accept that and I'm cool with it, but I'd like to be able to do like 4 of the backup jobs at once.
We'd like to be able to trigger replication by other backup servers at the end of a backup cycle so that the additional backup servers act as slaves to the master backup server rather than having 3 servers all across the world hit my clients server machines at once.
We can't set individual SSH ports for each backup source.
We'd really like to be able to make backups available to customers directly- our current structure and the structure made possible by rsnapshots (backupTier/clientName/serverName/directoryTree) doesn't allow for this. We'd like to have a system of categories for backup locations, so perhaps an FTP server can be made available for clients to access their backups (so something more similar to accessGroup/serverName/backupTier/directoryTree)
If you think that you can help us resolve these shortcomings while maintaining what we love about rsnapshot, please reach out to me on Discord: Airee#0001 Rsnapshot is written in Perl. Unless you want to offer to write us an entirely new backup system in another language, please know perl. Will pay fairly.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
