Why share-base level replication hit the high CPU resource than project-base level ?

Hi,

Our customer had used the Sun StorageTek 5320 with file replication.
they considered the retired 5320 and replace with ZFSSA.

I would like to suggest to set as share-based level replication because it be simular replication method (1:1 file system replication) of 5320.

I did test the two senarios with replication.

First scenario is one project and multi shares on this project. (30 file systems/shares).
Then it replicate 30 each shares with shared-based replication. / continuous replication.

The second scenario is one vs one (project vs share). there is one file system/share on one project. (total 30 file systems/30 projects).
Then it replicate 30 project-based replication / continuous replication.