Nice post. There doesn’t seem to be anything yet in the How-To category, but maybe you could fix that. 
Two things caught my eye.
Backblaze B2 confuses lots of people ever since application keys got introduced. I think the rule is to pair Key ID with Application Key, or pair Account ID with Master Application Key, but post suggests the pairing:
* Enter your Backblaze Account ID ( *e.g* . ##########################)
* Enter the Buckey ApplicationKey ( *e.g* . ###############################)
Removed account id to reduce confusion #3647 is an attempt to simplify things, but it might over-simplify. There’s some talk there about what works.
Duplicati only offers full backups ( not incremental or differential backups ). This will be an issue if your backups are large.
doesn’t reflect what happens, which is actually closer to incremental-forever, except with back-references instead of a (potentially breakable) chain of additional increments. Every file, old or new, is a set of blocks, and existing blocks can be used again in new files if the content is the same. That’s the deduplication part.
Taking a new “full” backup actually uploads only the changed blocks, and everything else just points to old. This makes what seems to be a full backup, with the backup overhead of an incremental. The inefficiency of having to rebuild restore files from potentially various bundles of blocks can slow down restores though.
Block-based storage engine is an easier overview.
How the backup process works is more technical.
Choosing sizes in Duplicati is done to fine-tune the algorithm to the use. Large backups sometimes suffer from too many blocks to track, too many little files on the remote, and so on. Some of this can be tuned…