I run a Tahoe-LAFS grid that spans a couple drives in my home machine and my office workstation. The majority of "archival" stuff I need to store that's large and relatively infrequently accessed (music, ripped DVD backups, hi-res TIFF scans of my artwork, RAW files for my photography) just lives in that grid. "Backups" are implicit. I even wrote my own music player that just streams out of Tahoe instead of off disk.

I actually had a (mostly full) 2TB drive go south just the other week and I was able to pull it out and replace it without any data loss. The rest of the grid was even still available while I replaced the drive. I put a new drive in, made a new Tahoe node on it, brought that node into the grid, then ran a "tahoe deep-check --repair" and it repaired and rebalanced all the files that had had shares on the broken drive.

My source code is vitally important, but small. I use git so I just always maintain a couple repos. Generally one on my workstation, one on our office git server, one on github, and there might be copies on my laptop, home machine, and a personal Rackspace server.

For my personal websites, I have a cron job that dumps data out nightly and drops it into Tahoe.