Seed bit-identical copy from local initial backup?

Support
#1

In order to reduce egress charges during a storage migration from the cloud, I’m curious if this would work:

(1) Create bit-identical copy-compatible storage on a NAS
(2) Run an initial complete backup job (which would be file-identical to my last revision stored in the cloud) to NAS storage
(3) Selectively copy revisions from cloud storage to NAS

The reason for doing this would be to only download the difference chunks for revisions I’m interested in. This doesn’t seem intuitive as I’m basically wanting to go “backwards” in time, but it would save me a substantial amount of egress if it would work.

#2

Sounds reasonable. Copy compatible bit- identical storage is initialized with the exact same keys, parameters, and seeds as the source, so there is a very high probability that you will end up with the same chunks, saving you from downloading them from the cloud.

#3

Ok, thanks for the encouraging response. I realized after I posted that I could just try the scenario rather than think about it. :slightly_smiling_face: I started my 13TB NAS population last night, and it should be ready a little later today. Once it’s done I’ll attempt to copy the latest cloud revision and see what happens. Will report back.

#4

Ok, the results are in. This method works very well.

The copy of the latest revision from cloud storage onto the NAS downloaded 6122 chunks (~60GB I’m surmising). Performing a diff between the two revisions I now have on the NAS shows no differences. Without having studied the internals of duplicacy, I am assuming the 6122 chunks represent the derivation path of my snapshot to its final state that I didn’t have by doing the initial backup the way I did. In other words, I now have the complete historical record that lead me to the current state, which is exactly what I wanted.

2 Likes
#5

Oh, one last note … in case someone is following in my footsteps - be sure to prune your “new” revision 1 after copying over the latest revision from the old storage. This will remove the “parallel universe” that was temporarily created during the process (and may save quite a bit of storage depending on how big your change sets are).