VVR has some intelligence of the filesystem (so can replicate just used blocks on an initial sync), but knows nothing about the application.  Also I am fairly certain that VVR does not read the current contents of a block before writing it to the SRL - i.e if a certain Block contains data "X" and you are writing the same "X" to it, then as VVR almost certainly does not read the current contents it does not know it is not actually changing the block.  For VVR to read each block, would put an extra overhead on VVR and almost certainly there will be some blocks changing during this maintenance (even if database is not open, Exchange will probably be writing blocks somewhere to denote progress of the maintenance), so VVR couldn't just assume that no blocks change during this process.   I don't understand what the point is of Exchange re-writing each block - what does this acheive?

Note VVR normally replicates continoulsy so would not do a full resync after DB maintenance but would replicate during the maintenance, but will probably get behind during this period and then catch-up from the SRL. 

Mike