Can I make a truly incremental cross-region snapshot copy of a an EBS volume?

Question

From reading the documentation, it would seem snapshot copies of the same volume are incremental. But that doesn't seem to be what I am seeing , performance wise, when I copy a snapshot of that volume to another region, i.e. from us-east-1 to us-west-2.

What I would really like to do would be create the initial snapshot from the volume directly into the West region, but that doesn't seem to be an option. So I have to first create a snapshot in east , a snapshot whose only purpose is to be copied to west.

So what I am doing now is

1. create snapshot of the east EBS volume in east. This does behave as if it is incremental, going much more quickly than a new original snapshot from a different volume of the same size.
2. Wait until this snapshot is complete by periodically polling the new snapshot id
3. copy the new snapshot from east to west. This does not seem to behave as it if were incremental, each one takes around as much time as the original snapshot, and the size is fairly constant.

What it seems like to me is that since it's not copying directly from the volume or the same snapshot each time, the snapshot copy to west doesn't know to be incremental.

Of course, I am also willing to accept that somehow I am not seeing what I think I am seeing , and the cross region snapshot copy are truly incremental. But considering it is taking up to 9 hours to do the copy every time it doesn't feel that way. And most of the documentation I have read seems to say its incremental from the same EBS volume, while when I do a describe on the snapshot copied out to west, it doesn't mention the original volume ID, but a dummy ID instead, which of course is a WAD.

-- background info on the nature of the data

Just some info since some might want to know how much data given the extended duration of our copy to west - the original EBS volume is the data volume of an ec2 instance used to store the backups of a source code control system generated by their proprietary backup tool.

The data volume has a few 'vendor snapshots' that are created with rsync under the hood , so each new 'vendor snapshot' has a lot of hard links between runs and a lot of commonality, with maybe 5% of the data changing between AWS snapshot copies.
The total EBS volume size is 3TB , of that 40% is unused space ever since we reduced the number of concurrent backups we keep.

Answer 1

I came back to this issue and dug into the documentation again. I don't know if there was an addition to the documentation, or I missed it initially, but there is now documentation that explains it. In order for an encrypted cross-region snapshot copy to be incremental, you must be using the default encryption key. After implementing this, our cross region copies went from anywhere between 4-12 hours to between 10 and 35 minutes. Much more manageable for achieving a reasonable RPO for our disaster recovery.