@bk3k
Wrong. These are databases. Source/replica is technically more precise than master/slave when describing what they're actually doing. But if MD's database of choice uses master/slave in its documentation, then by all means continue doing so.
It doesn't feel any faster (or slower) to me. Do you collect metrics that actually measure this objectively? E.g., median and 99%ile response times.
Can you stop using master/slave terminology? The industry has moved on. Major databases now use terms like source/replica.