ReFS - Working but under performing

[rosenstand]

Hi All

Sometime I hope to get one of these strange problems with my Backup so that I can write here, just because of all the awesome help and knowledge here are

And now, I got one of these strange problems.

Let me try to explain.

I have two identical physical servers running as my proxy and repositories.
Server 1 is located at my production site and got a DAS at 80TB connected to it.
Server 2 is located at my DRS site and got a DAS at 100TB connected to it.

Server 1 is running my daily backup and then copy these data to server 2 each day.

Both servers are running Server 2016, and the DAS storage is presented as a single disk, formatted in ReFS.
Veeam runs 14 jobs on Server 1, as the daily backup, and then there are 3 jobs to copy the data to Server 2.
4 jobs have a 7 days retention policy on Server 1.
5 jobs have a 14 days retention policy on Server 1.
4 jobs have a 31 days retention policy on Server 1.
All of this is working perfectly, and the data are at both sites. All good so far.

And now comes my strange problem.
At Server 1 I am using 54,3TB of the disc space when looking at the Disc Properties directly.
When I then look at the backup folder properties, I am using 86,9TB of disc space. - ReFS doing some magic.

At Server 2 I am using 30,7TB of the disc space when looking at the Disc Properties directly.
When I then look at the backup folder properties, I am using 74,1TB of disc space. - ReFS doing some serious magic.

So why do I think this is strange.
Because of my 3 copy jobs from Server 1 to Server 2.
Copy job 1 includes the 7 Days retention data at Server 1 and got a 14 days retention policy on Server 2.
Copy job 2 includes the 14 Days retention data at Server 1 and got a 31 days retention policy on Server 2.
Copy job 3 includes the 31 Days retention data at Server 1 and got a 60 days retention policy on Server 2.

So I am storing all of my data at the double of time of Server 2, and the ReFS magic are also better than on Server 1.
So what is wrong with my Server 1?

I can remember that I have been looking at ReFS drivers at some point back in time, after a Digest from Gostev, but I can't remember if I made any changes, and if so which one.

Any of you have some good ideas for me to try out, or have experienced something like this before yourself?

Thanks

[ejenner]

I think you'd have to know a lot more about what it's actually doing to become a master of it's ways. For instance, there are sizes of files ReFS will ignore because they're too small or too large... don't ask me what the figures are...

Just to put your mind at rest, here's some data from three of my ReFS volumes

Actual / Deduped / Compression

39.4TB / 76TB / 48%
28.4TB / 28.1TB / -1.06%
50TB / 143TB / 65%

[foggy]

Don't forget that backup copy job is always incremental, so the chain has only a single full with all subsequent restore points being incremental ones. Depending on the settings, regular backup job chains can have multiple fulls each, hence take more disk space and have less block clone savings if fulls are active.