Usability and Performance enhancements needed for very large filespace replication

There are 3 major problems currently with replication of very large filespaces.
1) The initial replication and ongoing replication of data for nodes with very large filespaces (100 million+ objects) and with a large amount of data (10+TB) works very well as long as the data on the source server is located in a disk pool.
Performance however drops to an order of magnitude or more slower if the data resides on tape. We even found that it was significantly faster overall to first allocate a massive disk pool, do a MOVE NODEDATA operation from tape pool to disk pool, and then run the initial replication to the target server. This is unacceptable and shows that the code was simply not optimized for the use case of tape.
2) Once replication is established, if REMOVE REPLNODE needs to be run on the node, the REMOVE REPLNODE command runs only in the foreground and seems to hang.
We believe the REMOVE REPLNODE command should run as a process (with the option of wait=yes) and report statistics of how many objects out of the total have been removed.
3) If a situation occurs where a replication sync needs to be performed between the source and target servers for the large node, there is a huge performance AND usability problem.
The time required simply to sync/confirm objects on the target seems to take many times longer than even the initial replication. Also, the replication sync process reports NO STATISTICS to show its progress or estimate of time to completion.