Answer
That answer has to come from business impact, not infrastructure preference. Teams should measure how long order entry, shipping, finance, customer support, or manufacturing can realistically pause before the interruption becomes expensive or unacceptable.
Once that tolerance is clear, the software discussion gets sharper. Buyers can judge replication lag, role-swap process, staffing requirements, and testing discipline against a real continuity target instead of a vague desire for less downtime.