I've got a reporting database that I'm mirroring to a secondary database using AlwaysOn sync replication. The nightly ETL process into the main database takes about an hour and generates about 60GB of transaction log space. The sync rate and queue seem very healthy with the queue at or near zero all the time. However, the redo rate seems stuck around 11MB/s and it takes a long time for the replica to get caught up -- about an extra hour after the hour long primary ETL.
The secondary has plenty of CPU and disk... cpu % is < 40% the entire time and disk response time is < 1ms (all flash target array).
are there knobs to turn here? any other information of interest?