In my experience, the quickest way to move whole volumes within a filer is vol copy.

There seem to be several concerns with using 'vol copy' in our situation:

1. The downtime - judging from our previous experience with 'vol copy', around 2 hours would be required to move 45 Gigs of data.

2. Someone has pointed out that 'vol copy' would not work in situation where the source and the destination volumes have different number of hard drives and different RAID group sizes.

If you need incremenal support, then SnapMirror also works great. The only downside to SnapMirror is it isn't free.

Someone already suggested using SnapMirror, and it seems it could be a viable option for us. Is there anything not mentioned in "Data Protection Guide" (related to SnappMirror procedure) that would be relevant to our situation?

But do you really need "infinite incrementals" I mean you should be able to do it with minimal downtime with only a few levels. Even if you did 1 incremental per day for a week before doing the final incremental with downtime, you should be able to fit within 10 levels of incrementals. How long do you intend to strech it out?

It has been reported in this tread that there are possible problems even with '-level 0' and '-level 1' backups with the Java version of NDMPcopy.

Robert