Timefinder Snap Query

Hi There

I have a customer who is taking SNAP copies of R2 devices in order to take copies of incremental updates to the R2 and present them to another host. The problem is that the incremental data refresh of the R2s is large each time and it is taking them a long time to complete this process before releasing the R2s for another refresh. It has been suggested that it might be faster for them to mount the R2s directly onto the host and do fs level copies.

My question is, will this be likely to actually save any time, given that the SNAP vdevs only contain pointers back to the source R2 devices. i.e does the SNAP process pull (copy) changed tracks through the SNAP tables in cache (or actual physical SAVE devs?) - in which case cutting out this step may save time. Or does it merely provide the pointers back to the source R2 devices to the host - in which case it would probably not save much time to copy directly from the RS source devs?

Any clarification on how Timefinder SNAPs work at the track level would be great.