"Cloud" just means "a server somewhere"
Pretty sure there are cloud vendors that support linux natively, but as long as you've got upload bandwidth to keep pace with the deltas, there shouldn't be any difference between J Random Cloud Vendor and J Random Server. I have a (linux) server at a mates house that keeps a versioned copy of my files, plus I keep offsites in the form of the portable hard drives.
If your dataset is <1TB and relatively static, no reason not to put it on SSD if that's within budget - should* be good for at least a couple of years, but for the same budget you could also buy a bevvy of 2TB platter-based drives and keep multiple offsites (one at work, one at your parents', that sort of thing) and cycle them - as long as you're reasonably careful, shock shouldn't be an issue. I call this Redundant Hodge-Podge of Offsite Cheap USB Backup External Drives, or RHOPOO-CUBED for short. I've got udev rules that kick off the backup script as soon as the backup drives are plugged in, and a cron that fires off reminder emails to bring in, say, RHOPOO-CUBED #4 if it senses it hasn't been backed up for a while.
If you're going to mail or courier them to a secure vault in an old salt mine, then you
could go SSD to eliminate the shocks, but IMHO money's better spent on some closed-cell foam blocks as these would come in handy for handing regular drives out to relatives as well.
The most reliable way of testing the viability of your backups is doing a test restore every so often - a backup where you haven't tested the restore is only a
potential backup and there's nothing worse than the false sense of security of
thinking you've got a backup. Here portable HDDs have a distinct advantage over optical and tape (depending on your backup method) that if you're doing file-based backups then you can generally just plug the drive in and clickety a few files - epitome of the KISS philosophy really.
* Canary-time again