Recommendations wanted: G, F, S cycles

[Catering1]

It's probably wrong to call this grandfather, father, son cycles but here's what I want to achieve.

1] Normal daily backups of all VMs to repository 1 with 14 instances retained.
2] Weekly deduped copy of backups from repository 1 to repository 2 with 7 instances retained.
3] Monthly copy from repository 2 to repository 3 with 12 instances retained.

In theory this approach should give us up to 12 months of backups/coverage.

The problem we have hit is that the monthly copy from repository 2 take too long to complete, with the result that some of the instances have been removed from the source repository and cannot be found/decompressed, so the whole copy action [job] fails.

Extending the retention period for repository 2 would cause the repository to run out of space.

For information> backups run on a proxy server [Windows 2012R2 and target for all three repositories is an iSCSI SAN with approximately 10Tb space on each repository. We are backing up 12 different Guest systems.

Can anyone suggest how one might best set things up so as to achieve the desired objective which does NOT involve making larger repositories, or altering existing repositories in any way.

[nmdange]

Change the monthly copy to copy from the normal backup.

[Shestakov]

nmdange is correct, you may try doing backup copies using the backup job as a source to see if it will give better performance.

some of the instances have been removed from the source repository and cannot be found/decompressed, so the whole copy action [job] fails.

This looks strange. Could you provide the error message text?
Thanks!