I am currently considering Data ONTAP 7-mode or ONTAP pre-9.4. In these versions, deduplication saving is directly counted into the latest snapshot. That is, if a volume has 50TB data with 20TB deduplicable, then after deduplication the volume become 30TB and the saved 20TB is counted into the newest snapshot if any. This is very intuitive since deduplication directly merges and deletes the duplicate blocks, and these blocks are therefore protected by the latest snapshot.
On the other hands, snapmirror uses snapshots to replicate volumes. When replication completes, common snapshots are kept by both source and desitination volumes. These snapshots are locked as current baseline, and will be automatically replaced by new baseline snapshots after a few snapmirorr updates.
The question is that if both snapmirror and deduplication are scheduled to run periodically, is there any chance to prevent the deduplication saving from being counted into a snapshot and then being replicated to the destination? Replacing local snapshot with snapvault is not considered because flexclone from snapshot is very handly and is used very frequently.
NetApp recommends to not run Dedup correlate to the same schedule as Snapmirror.
"SnapMirror creates a Snapshot copy before performing an update transfer. Any blocks in the Snapshot copy are locked and cannot be deduplicated. Therefore, if maximum space savings from deduplication are required, run the dedupe process before performing SnapMirror updates." - Page 44 chapter 10.5 https://www.netapp.com/us/media/tr-4015.pdf