Just speculating but it’s possible the block level is faster because it’s replicated over a dedicated and optimized SAN rather than (potentially) contending wit

Just speculating but it’s possible the block level is faster because it’s replicated over a dedicated and optimized SAN rather than (potentially) contending with normal network traffic. I assume the database state would only be crash consistent though.