There are many reasons why one might choose AOAG over FCI for a new highly available SQL Server cluster. However, there are issues with a large number of user databases due to the many threads needed to keep them in sync vs FCI shared storage block replication. Is there a threshold number of user databases where AOAG is no longer a viable architecture? What about simply scaling the compute to compensate?
